JP2007265333A - Operation restoration support system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To map system configuration and environmental configuration in an operation application management system to other system at a remote place, to automatically continue application without changing an operation definition even on a machine with different host name from an original system when a failure occurs in the original system and quickly and easily restart an operation. <P>SOLUTION: This operation restoration support system is constituted of the operation application management system constituting of a manager and a plurality of agents and a standby system operation application management system which acts for functions of the operation application management system when the failure occurs. Each host of an active system and each host of a standby system are mapped and when the failure occurs in the host of the active system, an operation is alternately executed by the mapped host of the standby system. When configuration change occurs between the active system and the standby system, mapping is set so that a load of the host of the standby system may be evenly distributed to update configuration management information. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a business recovery support system that enables a business recovery in the event of a system failure in a business operation on an electronic computer to be quickly performed at a remote location.

  In a business operation management system using an electronic computer, there is a risk of system failure due to various factors. For this reason, the construction of a response system in the event of a failure has been implemented for a system that is operating in a company's core business.

  In recent years, not only backup of important data but also a response system in the event of a failure by a (cluster) system having a redundant configuration has been constructed. In addition, even in the event of a wide-area disaster, even if the entire site is destroyed by remote disk mirroring or data replication, it is now possible to reconstruct the system at another site and resume operations.

  In the computer system having the redundant configuration, the system failure automatic recovery processing is integratedly managed by using a failure monitoring apparatus, and the recovery operation is automatically performed according to the type of the failure that occurred. By making it feasible, the man-hours for system maintenance are reduced.

Further, as a method of a remote data shadowing system for copying data to a remote place, for example, a method described in Patent Document 2 below is known.
JP 2003-114811 A JP-A-7-244597

  However, while the conventional technology can automatically transfer and save necessary data to remote locations by setting, it is quick to rebuild the system configuration at the time of failure and to recover from the failure. And it's hard to say that it can be done easily. In order to reconstruct the system configuration, it is necessary to prepare the same system configuration as the original system, or it is necessary to reconstruct an environment where the same operations as the original system can be performed manually. This is because there are circumstances.

  The technique disclosed in Patent Document 2 is a technique for transferring data to a remote location, and as it is, the system cannot be restarted at a remote location. Further, although the technology disclosed in Patent Document 1 can automatically recover system components according to the type of failure, the computer OS is premised on a redundant configuration (cluster) system. It cannot handle business recovery on another computer with a completely different location. For example, if business operation management is performed in a manager / agent configuration, the agent machine information for executing the business is included in the business definition information. To execute the same business on another system, It must be possible to operate with the host name (computer name) specified in the business definition. For this purpose, it is necessary to prepare the same machine environment as the original system. Or, when restarting a task on another system, change the communication setting information between machines so that each task can operate without problems, or change all task definitions so that it can operate on another system. Don't be.

  The present invention maps a system configuration and environment construction in a business operation management system to, for example, another system at a remote location, and when a failure occurs in the original system, the original system is different from the host name (computer name). The purpose of this technology is to provide a technology that enables automatic operation continuation without changing the business definition even on different machines, and that the business can be resumed quickly and easily.

  In order to achieve the above object, the present invention comprises a manager host and an arbitrary number of agent hosts managed by the manager host, and the agent host responds to a request from the manager host. In the business operation management system to be executed, when a failure occurs and the business operation management system stops, a manager host that performs alternative operation and an arbitrary number of agent hosts managed by the manager host are provided. Establish a standby business operation management system. In addition, a configuration management table that defines in advance which manager host and which agent host of the standby business operation management system will perform business substitution when a failure occurs in the manager host and agent host of the business operation management system is prepared. In addition, the correspondence between the host names of the manager host and agent host in the business operation management system and the IP address that identifies the host of the standby business operation management system that substitutes for those hosts is stored. Prepare a host name resolution table. When it is detected that the manager host of the business operation management system has stopped, the manager host of the standby business operation management system, which replaces the stopped manager host, is assigned from the configuration management table, and the business is continued. Further, when detecting that the agent host of the business operation management system has stopped, assigns an agent host of the standby business operation management system as a substitute for the stopped agent host from the host name resolution table, Let it continue.

  When a new agent host is added on the manager host of the business operation management system, the configuration management information is automatically updated, the standby business operation management system is notified, and the standby business operation management system In the manager, when an agent added to the business operation management system stops, the work in the business operation management system is automatically distributed to each agent in the standby business operation management system, and executed as an alternative. It is advisable to set the configuration definition so that In addition, when a new agent host is added on the manager host of the standby business operation management system, the configuration management information is automatically updated, and the largest number of existing agent hosts are replaced. It is advisable to search for a host that is defined and to perform configuration definition settings so that the load is automatically distributed to the newly added agent host.

  According to the present invention, the system configuration and environment construction in the business operation management system are automatically mapped to another system at a remote location, for example, on a machine having a host name (computer name) different from that of the original system. However, automatic operation can be continued without changing the configuration information such as the host name or the business definition, and the business can be resumed quickly and easily even if a large-scale failure occurs.

  Hereinafter, an embodiment for carrying out the present invention will be specifically described with reference to the drawings.

  FIG. 1 is a block diagram showing a schematic configuration of a system according to an embodiment of the present invention. The remote business recovery support system of this embodiment includes an active system 1 and a standby system 2. The active system 1 includes a base 3 and a base 4. The base 3 is provided with a manager 6 that has business definition information and controls business execution, and agents 7 and 8 that execute actual business in response to a request from the manager 6. The base 4 is provided with an agent 9 that executes an actual job in response to a request from the manager 6. The standby system 2 includes a base 5. At the base 5, if the manager 6 of the active system 1 fails, the manager 10 that takes over the work and the agent 7, 8, or 9 of the active system 1 Agents 11 and 12 that take over and execute are provided. Each base is connected by a network 13 (for example, WAN (Wide Area Network)). Each of these managers and agents may be considered as one machine.

  FIG. 2 is a block diagram showing a mapping configuration between the active system and the standby system. Mapping refers to a definition that associates which machine in the standby system takes over work when a failure occurs in each machine in the active system. In the example of FIG. 2, the manager 21 of the active system (6 in FIG. 1) and the manager 25 of the standby system (10 in FIG. 1) are mapped. If the manager 21 stops due to a failure, the manager 25 takes over and continues the business. Similarly, agent 22 (7 in FIG. 1) and agent 24 (9 in FIG. 1) are agent 26 (11 in FIG. 1), agent 23 (8 in FIG. 1) is agent 27 (12 in FIG. 1), Each is mapped. When any of the agents stops due to a failure, the manager requests the agent of the standby system to execute the business. The agent 26 is mapped to a plurality of agents 22 and 24 in the execution system. When both the agent 22 and the agent 24 are stopped, the business for two agents is executed by the agent 26.

  FIG. 3 is a configuration diagram of a configuration management table that manages the mapping configuration of FIG. This configuration management table defines the host name 31 in the active system, the host name 32 of the standby system mapped to the host in the active system, and the IP address 33 of the host in the standby system To do. Also, a type 34 indicating whether each host is a manager or an agent, and a standby activation mode 35 in the case of a manager are defined. This configuration management table is created in advance by an administrator and stored on the managers 21 and 25 of the active system and the standby system. The data update of the configuration management table is performed by the active manager 21 based on an instruction from the administrator. When the configuration management table in the active manager 21 is updated, the updated data is transferred to the standby manager 25, and the configuration management table in the standby manager is updated in the same manner.

  When a job switch to the standby system occurs, the data in this configuration management table is read. If the start mode is "Automatic", the business data transferred to the standby system after starting the business operation management system on the standby system Automatically resumes work based on When the start mode is “manual”, after starting the business operation management system in the standby system, the business is suspended. In this case, the execution status of the business is confirmed with an execution log in the business data, and the business is manually restarted. The process switching process will be described in detail later.

  FIG. 4 is a block diagram showing transfer of business data between the active system and the standby system. A database 45 storing business definition data, business execution status data, business execution result data, configuration management data, and the like is connected to the manager 43 (6 of FIG. 1) of the active system 41 (for example, 1 of FIG. 1). ing. When an update occurs in the database 45, the data in the database 45 is reflected in the database 46 connected to the manager 44 (10 in FIG. 1) of the standby system 42 (2 in FIG. 1) by the replication function of the database software. The Thereby, during business operation, the same data as the database 45 in the execution system 41 is held in the database 46 in the standby system 42.

  FIG. 5 is a configuration diagram of a host name resolution table that is a data table for host name resolution. A flag area 53 for storing a host name 51 and each IP address 52 in the execution system and a flag indicating whether or not the host is currently in use (in the example of FIG. 5, 0 indicates that it is in use and 1 indicates that it is not in use) And the priority 54 of the IP address is stored. When the running system is operating normally, out of the records where the host name 51 is the host name of the running system that is operating normally, the running system where the IP address 52 is operating normally The flag area 53 of the record that is the IP address of the current host is 0 (in use), and for other records with the same host name, the IP address 52 is the host of the standby system (the host of the above-mentioned execution system) And the flag area 53 is 1 (unused). In communication processing between the manager and the agent, address resolution is performed according to the information in the host name resolution table. When resolving the address of the execution agent set in the job definition information, the manager inputs a record in which the host name 51 is the execution agent and the flag area 53 is 0 (in use). Communication is performed using the IP address 52 set in this record. In the communication process to the agent, the manager's host name and IP address are reported to the agent. Also at this time, the information of the host name resolution table is referred to. A record in which the IP address 52 is the same as that of the own host and the flag area 53 is 0 (in use) is input. The agent is notified of the host name 51 and IP address 52 set in this record. Communication from the agent to the manager uses this host name and IP address.

  For example, in the first record of the host name resolution table in FIG. 5 (assuming that the running system is operating normally), the host name 51 is “manager 21”, the IP address 52 is “1.0.0.1”, The flag area 53 is “0”, and the priority 54 is “1”, which indicates a host of an active system that is currently operating normally. In the next record, the host name 51 is “manager 21”, the IP address 52 is “2.0.0.1”, the flag area 53 is “1”, and the priority 54 is “2”. This record is mapped to the active host “1.0.0.1”, and indicates a standby host that takes over the “manager 21” when a failure occurs in the active host. If a failure occurs on the executing host, enter the manager 21 record, change the flag area 53 of the executing host record to 1 (unused), and set the standby host record flag area. Change 53 to 0 (in use). In the subsequent communication processing, since the record of the standby host is in use, “2.0.0.1”, which is the standby host, is used as the IP address of the host name “manager 21”.

  The priority level 54 indicates the priority level when a failure occurs and the standby host takes over (assuming that the lower the numerical value, the higher the priority level). When a failure occurs in the host of the active system, the IP address 52 with the highest priority 54 is acquired from the IP addresses 52 of the standby system having the same host name 51 as that host, and the standby host of that IP address is acquired. In step 3, take over the host of the active system in which the failure occurred. At this time, the flag area 54 of the record indicating the host of the standby system that has taken over the job is updated from 1 (unused) to 0 (in use).

  FIG. 6 is a flowchart showing an outline of the recovery process in the standby manager when the manager of the active system stops.

  The standby manager periodically communicates with the active manager (the active manager mapped to the standby manager as described in FIG. 2) to check whether the active manager is running. . If the active manager is running, the process is terminated, and if it is stopped, the process proceeds to the next step (step 601). If the active manager has stopped, the standby manager reads the configuration management table described in FIG. 3 (step 602). The record of the same host name 51 as the manager host name in the host name resolution table of FIG. 5 is input, the flag area 53 of the record of the executing host is set to 1 (unused), and the flag area 53 of the record of the standby host is set to 0 (In use). Thereafter, the IP address of the manager host notified from the manager in the communication processing is the IP address of the standby host (step 603). It is checked whether the activation mode 35 of the record whose execution host name 31 in the read configuration management table is the stopped manager is automatic activation or manual activation (step 604). In the case of automatic startup, the standby system manager starts the business system and resumes the business (step 611).

  In the case of manual activation at step 604, the first agent among the agents of the active system registered in the active host name 31 of the configuration management table of FIG. move on. Referring to the host name resolution table shown in FIG. 5, among the records having the same host name 51 as the host name of the processing target agent, the record with the highest priority 54 (the one with the smallest value) is set as the first target record. (Step 605). Next, it is checked whether the agent identified by the IP address 52 of the target record is stopped or operating (step 606). If it is stopped, the flag area 53 of the target record is set to 1 (unused) (step 607), and the record having the next priority of the target record in the host name resolution table of FIG. Search is made for those having the same host name 51 as the host name (step 608). If there is, the record is newly set as a target record, and the process returns to Step 606 (Step 609). If there is no record having the next priority in step 609, it means that there is no active host that realizes the agent to be processed, so an error log is output (step 610), and the process is terminated.

  If the agent specified by the IP address 52 of the target record in step 606 is operating, the flag area 53 of the target record is set to 0 (in use). When the manager resolves the IP address from the host name of the job execution destination agent during subsequent communication, this record is referred to, and the job execution destination is switched to the standby agent (step 612). It is determined whether there is an active agent that has not yet been processed among the agents registered in the active host name 31 of the configuration management table of FIG. 3 (step 613). If there is an unprocessed agent, the agent returns to step 605 as the new agent to be processed. If there is no unprocessed agent, it means that the processing has been completed for all the agents registered in the active host name 31 in the configuration management table of FIG. 3, so the business system is started in standby mode (step 611). .

  The activation of the business system in the standby mode is a process of starting the system and waiting while suppressing execution of the business. In the case of automatic activation, the process proceeds from step 604 to step 611. In this case, the business system is activated in the automatic mode and the business is immediately resumed. Further, for the flag area 53 of the manager record in the host name resolution table of FIG. 5, the flag area 53 of the stopped active manager is set to 1 (unused), and the flag area 53 of the standby manager that has taken over the task is set. Although it is necessary to perform a process of 0 (in use), this process is performed in the system startup process in step 611.

  As described above, in the manager recovery process, in the case of automatic startup, the business system is started immediately and the business is restarted. In the case of manual startup, the status of each agent is checked, the configuration is changed, and the business is changed. Start the system and put it in a business restart standby state.

  Checking the operation / stop of each agent and changing the configuration in the case of automatic startup is the same as in the case of manual startup when the business is actually restarted and a business execution request is made to the corresponding agent. Processing (steps 604 to 610, 612, 613) is performed.

  FIG. 7 is a flowchart showing an outline of recovery processing when the agent of the execution system stops. This process is started in the manager when the manager makes a business execution request to the agent and the agent is stopped.

  First, a record in which the IP address 52 matches the IP address of the stopped agent is searched from the host name resolution table shown in FIG. 5 (step 701). The flag area 53 of the record of the searched stop agent is set to 1 (unused) (step 702). Next, the host name resolution table in FIG. 5 is searched for a record having the same host name 51 as the stopped agent and the next highest priority 54 of the stopped agent (step 703). If there is no record with the next highest priority, it means that there is no standby agent that takes over the stopped agent, so an error log is output and the process is terminated (steps 704 and 708). If there is a record of the next priority, it is checked whether the agent identified by the IP address 52 of that record is operating (step 705). If it is operating, the flag area 53 of that record is set to 0 (in use) ). When the manager resolves the IP address from the host name of the job execution destination agent during subsequent communication, this record is referred to, and the job execution destination is switched to the standby agent (step 706). A business execution request is sent to the agent of the IP address (step 707). If the agent is stopped (step 705), the process returns to step 702.

  As described above, in the agent recovery process, when an agent is stopped, a task is executed by other active agents that are mapped (assigned in descending order of priority). The same processing is performed for notification of the business execution result from the agent to the manager, and notification is given to the other active managers that are mapped.

  FIG. 8 is a flowchart showing an outline of the configuration change process when an agent is added to the active system.

  The active manager adds a new agent (step 801), and adds a record storing only the host name 31 of the active system to the configuration management table of FIG. 3 (step 802). As described with reference to FIG. 4, if there is an addition / change in the configuration management table in the active manager, the standby manager is notified to that effect. The standby manager adds the record in the same manner to the configuration management table of FIG. 3 held in its own apparatus, and reads data with the current flag 104 ON from the agent list table shown in FIG. 10 (step 803). ).

  Here, the agent list table in FIG. 10 will be described. The agent list table is a table held by the standby manager. The agent list table has an agent ID 101, an agent host name 102, an IP address 103, a current flag 104, and a mapping agent number 105. The agent ID 101 is a serial number here. The agent host name 102 describes the standby host name. The IP address 103 is the IP address of the corresponding agent host name 102 host. The current flag 104 indicates a standby agent to be assigned next. That is, when a standby agent to be mapped becomes necessary, an agent whose current flag 104 is ON is used. Among the agents registered in the agent list table, only one has the current flag 104 ON, and the current flags 104 of the remaining agents are OFF. When an agent whose current flag 104 is ON is assigned as a standby agent, the current flag 104 of that agent is turned OFF, and the current flag 104 of the agent whose agent ID 101 is the next number is turned ON. After the last agent in the agent list table, the agent ID 101 returns to the first agent of 1. The number 105 of mapping agents indicates the number of active agents mapping the agent as a standby system.

  Returning to FIG. 8 again, after step 803, the standby manager notifies the execution manager of the host name 102 and IP address 103 of the agent whose current flag 104 is ON (step 804). It is turned OFF and the current flag 104 of the record with the next ID is turned ON (step 805). The active manager updates the configuration management table in FIG. 3 and the host name resolution table in FIG. 5 based on the agent information (host name and IP address) notified in step 804 (step 806). The information updated here is reflected in the standby manager as shown in FIG.

  As described above, when a new agent is added, mapping with the standby agent is automatically performed and reflected in the database. This configuration change function can be executed only manually by setting.

  FIG. 9 is a flowchart showing an overview of the configuration change process when an agent is added to the standby system.

  The standby manager adds a new agent (step 901), and adds a record to the agent list table of FIG. 10 (step 902). The record having the maximum number of mapping agents 105 is input from the agent list table, the value is counted down by 1, and the number of mapping agents 105 of the record added in step 902 is set to 1 (step 903). Next, the agent name record in which the number of mapping agents is counted down in step 903 is read from the configuration management table in FIG. 3, and the change of data is notified to the executing system manager (steps 904 and 905).

  The active manager updates the agent to be mapped in the configuration management table based on the notified agent information, and updates the corresponding IP address in the host name resolution table (step 906).

  As described above, when a new agent is added in the standby system, the mapping with the active system agent is updated, and the agents are automatically distributed to the standby system to which multiple agents are mapped. . This configuration change function can be executed only manually by setting.

  In the above embodiment, the manager of the standby system detects the stop of the manager of the active system and performs the processing of FIG. 6, but the function of detecting the manager stop, the function of starting the system, etc. You may perform these processes with another apparatus provided with. The process in FIG. 7 is the same.

The block diagram showing the structure of one Embodiment of this invention Block diagram showing the mapping configuration between the active system and the standby system Data structure diagram for managing mapping structure Block diagram showing transfer of business data between active and standby systems Data structure diagram for host name resolution Flow chart showing the outline of recovery processing when the manager of the active system stops Flowchart showing an overview of recovery processing when an agent in the running system stops Flowchart showing the overview of configuration change processing when an agent is added to the active system Flowchart showing an overview of configuration change processing when an agent is added to the standby system Data structure diagram for managing the agent list

Explanation of symbols

  1 ... active system, 2 ... standby system, 3, 4, 5 ... base, 6 ... execution system manager, 7, 8, 9 ... active system agent, 10 ... standby system manager, 11, 12 ... Standby system agent, 13 ... Public network, 21 ... Active system manager, 22, 23,24 ... Active system agent, 25 ... Standby system manager, 26,27 ... Standby system Agent, 31 ... Execution host name, 32 ... Standby host name, 33 ... Standby IP address, 34 ... Type, 35 ... Startup mode, 41 ... Execution system, 42 ... Standby system, 43 ... Execution system Manager, 44 ... Standby system manager, 45 ... Execution system business operation management system DB, 46 ... Standby system business operation management system DB, 51 ... Host name, 52 ... IP address, 53 ... Flag area 54 ... Priority, 101 ... Agent ID, 102 ... Agent host name, 103 ... IP address, 104 ... Current flag, 105 ... Number of mapping agents.

Claims (1)

In a business operation management system comprising a manager host and an arbitrary number of agent hosts managed by the manager host, the agent host executes an actual business in response to a request from the manager host. A business recovery support system that supports business recovery when it occurs,
In the case where the business operation management system is stopped, a standby business operation management system including a manager host that performs an alternative operation and an arbitrary number of agent hosts managed by the manager host is provided.
Means for storing in advance a configuration management table that defines which manager host and which agent host of the standby business operation management system perform business substitution when a failure occurs in the manager host and agent host of the business operation management system When,
Host name resolution table that stores the correspondence between the host names of the manager host and agent host in the business operation management system and the IP address that identifies the host of the standby business operation management system that substitutes for those hosts Means for storing;
Means for allocating a manager host of the standby business operation management system as a substitute for the stopped manager host from the configuration management table when detecting that the manager host of the business operation management system has stopped; When,
When it is detected that the agent host of the business operation management system has stopped, an agent host of the standby business operation management system that substitutes for the stopped agent host is assigned from the host name resolution table, and the business is continued. A business restoration support system characterized by comprising means.

Images