JP3409895B2 - Load distribution method and information processing system - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a load balancing method for an information processing system having a plurality of processing systems. In an online system in which a plurality of terminals access a host, it is desirable to reduce the load on the host as much as possible in order to improve the responsiveness of access.

On the other hand, in an information processing system of a centralized processing system, particularly in a system which supports highly public works, a main resource including a CPU is duplicated in order to avoid a system down, and normally, The duplex method is adopted as the method.

[0003]

2. Description of the Related Art In a duplex system, when the resources of the active system are operating normally, the resources of the standby system can be used for other purposes within a range that does not hinder the hot standby operation. Can improve the economic efficiency of the entire system. That is, distributed processing can be performed by the active system and the standby system.

Conventionally, various processes (jobs) have been performed in advance.
A load balancing method has been proposed in which a system (active system or standby system) for executing them is fixedly set, and one and the other processing systems are each only responsible for a specific process (Japanese Patent Laid-Open No. Hei 10 (1999) -242242). 4-52953).

[0005]

However, in the conventional load balancing method, the load (job) is distributed regardless of the operating state of the system, so that the system is in a low load state and any of a plurality of processing systems is used. Even when each job can be executed without delay, the jobs are handed between the processing systems, which may increase the load.

For example, in an online system, when the host adopts the duplex system, if the terminal is allowed to access a plurality of processing systems in the host at the same time, each processing system will perform duplicate operation. Therefore, in this case, it is common to configure the host so that only one processing system accepts access from the terminal, and requests processing to another processing system by communication between the systems as necessary. Is. In an online system having such a host, if the processing system responsible for accepting access is in a high load state, it is possible to avoid the occurrence of an overload state by handing over a specific job to another processing system. On the other hand, if the processing system responsible for accepting access is in a low-load state, an extra load (communication, etc.) due to handing over the job to another processing system is generated as compared with the case where this processing system itself executes the job. It will be added to each processing system.

In view of the above problems, it is an object of the present invention to suppress the increase of unnecessary load and to operate the information processing system more efficiently.

[0008]

In order to solve the above-mentioned problems, the method according to the invention of claim 1 transfers jobs as necessary by mutual communication as shown in FIGS. 1 and 3. At least one having a plurality of processing systems 10 and 20
In the information processing system 1 configured to enable hot standby operation by another processing system when a failure occurs in one of the processing systems, the processing system to be executed is determined according to the system operating state for each of various jobs. provided in advance a system distribution table TA for sorting Te,
Wherein the processing systems 10 and 20, monitors the system operation state, when a job is given, with reference to the system distribution table TA in light of the system operation status at that time,
For a given job, in normal operation , the processing system 10, 2 of the own system or another system indicated by the system allocation table TA
Execute the first system allocation process that allocates 0
During the standby operation, another processing system with the failure executes
The second process as an alternative process of the first system allocation process
System allocation processing is executed, and the second system allocation processing and
Handing over jobs between the first system distribution process of its own
Is performed together with the first system allocation processing of itself .

According to a second aspect of the present invention, when the system operating state is a low load state, the processing system to be executed is fixedly assigned to each of the various jobs, and the system operating state is a high load state. When in the state,
It is a method of dynamically allocating a processing system to be executed for each of the various jobs.

In the method according to the third aspect of the present invention, each of the processing systems 10 and 20 executes a job received from another processing system in its own system without referring to the system allocation table TA. Is.

An information processing system according to the invention of claim 4
Pass jobs as needed through mutual communication.
At least one processing system having a plurality of processing systems
When a failure occurs in the
It is an information processing system that is configured so that
Process to be executed for each of the various jobs
System allocation for allocating the system according to the system operating status
A table is provided for each processing system.
A means of monitoring work status, and when a job is given,
Based on the system operating status at that time,
Means to refer to the table and the resources provided during normal operation.
For the job, the own system or other shown in the system distribution table
Executes the first system allocation process that allocates the system processing system
Measures and other failures that occurred during hot standby operation
Other processing systems for the job given to
As an alternative process of the first system allocation process that was being executed
The second system allocation process is executed, and the second system allocation process is executed.
And receiving jobs during its own first system allocation processing.
The process of handing over is called the first system allocation process of itself.
And means for doing so are provided.

[0012]

In normal operation in which all the processing systems 10 and 20 can operate, each processing system 10 and 20 executes the system allocation process, and regards the job given to itself as the processing system which should execute it. The system or the other system is distributed according to the system distribution table TA. Then, for a job to be executed by another system, it is passed to another processing system through communication between the processing systems, and for a job to be executed by the own system and a job received from another processing system, it is sent to the other system. To run in.

In the system allocation table TA, for example, for each system operating state classified by the resource usage rate,
Various jobs are associated with processing systems that should execute them.

As a result, various jobs are executed in the form of distributed processing by the processing systems 10 and 20, and the processing system that executes each job is changed in accordance with the system operating state.

[0015]

1 is a block diagram showing a functional configuration of a host computer system 1 to which a load balancing method of the present invention is applied during normal operation, and FIG. 2 is a diagram showing an example of contents of a system allocation table TA. 3 is a block diagram showing a functional configuration of the host computer system 1 of FIG. 1 during hot standby operation.

In FIG. 1, a host computer system 1 is a central part of an online network constructed as a corporate network, for example, and has two processing systems 1.
It has 0 and 20.

The processing systems 10 and 20 have the same hardware resource and software resource configurations, and are capable of performing mutual hot standby operation in which when one fails, the other immediately complements it. In the host computer system 1, a job control library (JCL) 41 that is a set of job control statements and a program library 4 that is a set of various programs (modules)
2 and specific software resources such as the business database 43 are shared by the processing systems 10 and 20.

During normal operation in which both the processing systems 10 and 20 operate normally, information processing for supporting various jobs is performed in the form of distributed processing by the processing systems 10 and 20. That is, basically, the processing systems 10 and 20 are not distinguished into the active system and the standby system (standby system), and jobs are given to both of them in parallel. However, one processing system is responsible for accepting access by the online terminal 3. In the example of FIG. 1, a job related to access by the online terminal 3 is given to the processing system 10 via the network communication control processing device 2.

Each processing system 10, 20 is provided with a system distribution processing program 33, an operating state monitoring program 34, and a system distribution table TA in order to effectively utilize resources by load distribution.

In the processing system 10, jobs such as sales reports and inquiries are given to the system distribution processing program 33 from the online processing program 31 that accepts access from the online terminal 3. Also, from the batch processing program 32 to the system distribution processing program 3
For example, a job such as creating a daily sales table is given to No. 3.

The system allocation processing program 33 inquires of the operating status monitoring program 34 about the operating status of the host computer system 1, and refers to the system allocation table TA based on the result. Operating status monitoring program 34
Monitors the load status of predetermined resources in its own system represented by the CPU, and periodically communicates with the operating status monitoring program 34 of the other system to check the operating status of the entire host computer system 1 (system operating status). ).

In the system allocation table TA, as shown in FIG. 2, the processing system for executing each of the various jobs is registered in advance by classifying it into a low load state and a high load state. The low load state means, for example, each processing system 1
The resource utilization rates of 0 and 20 are both below a certain value, and the high load state is a state where the resource utilization rate of at least one processing system exceeds a certain value.

In the system distribution table TA of this embodiment, the contents of the portion corresponding to the low load state are fixed, and the contents of the portion corresponding to the high load state are variable. That is, the system administrator can display or print out the job being executed and the system operating state by operating a console (not shown), and based on the result, the registered contents of the portion corresponding to the high load state can be displayed. It can be changed in real time and the processing system can be dynamically assigned to each job. When the contents of the system allocation table TA of one processing system are changed, the contents of the system allocation table TA of the other processing system are also automatically changed.

Returning to FIG. 1, the system distribution processing program 3
3 is a system allocation table T for a given job.
The own system (processing system 10) or other system (processing system 20) indicated by A is distributed.

That is, when the job is distributed to its own system, the job (strictly speaking, a parameter that defines the content of the job) is delivered to the job activation processing program 35. The job activation processing program 35 refers to the job control library 41 and loads necessary modules from the program library 42 into the work area of the CPU. As a result, the job is executed in the own system.

When the job is distributed to another system, the job is delivered to the communication control program 36. The communication control program 36 is the communication control program 3 of the processing system 20 via the inter-system communication block 37 provided in each of the processing systems 10 and 20, and the communication link 51 connecting them.
6 and requests the other system to execute the job.

On the other hand, the processing system 2 requested to execute the job
At 0, the job is given from the communication control program 36 to the system distribution processing program 33. The system allocation processing program 33 of the processing system 20 immediately delivers the job received from the processing system 10 (another system viewed from the processing system 20) to the job activation processing program 35 without referring to the system allocation table TA.

Job activation processing program 3 of processing system 20
Reference numeral 5 refers to the job control library 41 and loads a necessary program from the program library 42 into the work area of the CPU. As a result, the job is processed by the processing system 20.
Will be executed in.

Further, the system distribution processing program 33 of the processing system 20 also refers to the system distribution table TA for the job given from the batch processing program 32 after inquiring the system operation status to the operation status monitoring program 34. Distribute the system or other system.

Thus, the host computer system 1
Then, the jobs are delivered between the processing systems 10 and 20 according to the system operating state to distribute the load. Note that the processing system that executed the job in response to a request from another system notifies the requesting processing system of the execution result for the job related to online processing, and the job attribute file for the job related to batch processing. By referring to this, whether or not the result notification is necessary is checked, and if the notification is necessary, the execution result is written in the shared file. The processing system of the request source reads the execution result of the job related to the batch processing from the shared file in a timely manner.

Next, the hot standby operation of the host computer system 1 will be described with reference to FIG. When a failure occurs in one of the processing systems (for example, the processing system 10) that was in charge of online processing during normal operation and the operation stops, the communication for confirming the mutual operation performed between the processing systems also stops. .

In the other processing system 20, the online processing program 31, the batch processing 32B, and the system distribution processing program are executed in order to complement the function of the processing system 10 when the communication interruption is detected by the inter-system communication block 37. 33B is activated. System distribution processing program 33
By starting B, the first and second system allocation processing programs 33, 33B are executed in parallel in the processing system 20.

Then, the communication control processing device 2 includes the processing system 1
Instead of 0, online processing program 3 of processing system 20
The communication connection state is switched so that the job from the online terminal 3 is given to 1.

In this state, when a job is given from the similar batch processing program 32 to the first system distribution processing program 33 which is being executed continuously from the time of the general operation, the system distribution processing program 33 is described above. In this way, the system allocation table TA is referenced to allocate the own system or another system to the job.

At this time, when the other system is distributed, the system distribution processing program 33 is the communication control program 36.
Instead, the job is delivered to the second system distribution processing program 33B. Upon receiving the job, the system allocation processing program 33B immediately transfers the job to the job activation processing program 35 without referring to the system allocation table TA. As a result, a job executed in another system during normal operation is executed in the own system (processing system 20).

On the other hand, the system allocation processing program 33B is
When a job is given from the online processing program 31 or the batch processing program 32B, the operating status monitoring program 34 is inquired about the system operating status, the system allocation table TA is referred to based on the result, and the system is assigned to the job according to the system allocation table TA. Or, allocate other systems.

At this time, when the other system is distributed, the system distribution processing program 33B delivers the job to the first system distribution processing program 33. Upon receiving the job, the first system allocation processing program 33 immediately delivers the job to the job activation processing program 35, as in the case where the job is received through intersystem communication during normal operation.

That is, during hot standby operation, as a substitute for the system allocation processing program 33 of the processing system in which the failure has occurred, the
System allocation processing 33B is executed, and the processing systems 10 and 20
Instead of passing jobs between jobs, jobs are passed between the first and second system allocation processing programs 33 and 33B. As a result, the functions of the host computer system 1 seen from the outside are almost the same as those in normal operation.

In the above embodiment, the system distribution processing programs 33 and 33B and the operating state monitoring program 34
Are not limited to pure programs, and may be firmware, for example.

In the above-described embodiment, the system allocation table TA is stored in the DASD together with the program library 42 and the like.
Provided in the (direct access storage device), the processing system 10, 20
May share one system allocation table TA.

Further, the contents of the system allocation table TA can be variously changed. For example, the system operating state can be divided into three or more, and the processing systems to be distributed can be registered in association with jobs for each classification.

In the above-mentioned embodiment, the system administrator has explained that the contents of the system allocation table TA are changed as necessary, but the processing systems 10 and 20 recognize each other's load condition and move. The contents of the system allocation table TA may be automatically changed so as to distribute the load.

The present invention can also be applied to a system having three or more processing systems.

[0044]

According to the present invention, the processing systems are distributed to the jobs according to the system operating status such that the other processing systems are requested to execute the jobs only when the processing systems are overloaded. Since there is no problem even if it is executed in the own system, it is possible to suppress the unnecessary load increase due to handing over the job to another system, and to operate the information processing system more efficiently. it can. Also obstacles
Hot stamper regardless of human operation.
B. Operation can be started.

According to the invention of claim 3, the processing can be speeded up while avoiding the overlapping operation .

[Brief description of drawings]

FIG. 1 is a block diagram showing a functional configuration of a host computer system to which a load balancing method of the present invention is applied during normal operation.

FIG. 2 is a diagram showing an example of contents of a system allocation table.

3 is a block diagram showing a functional configuration of the host computer system of FIG. 1 during hot standby operation.

[Explanation of symbols]

1 Host Computer System (Information Processing System) 10, 20 Processing System TA System Distribution Table 33 System Distribution Processing Program (First System Distribution Processing) 33B System Distribution Processing Program (Second System Distribution Processing)

Continuation of the front page (56) Reference JP-A-5-216842 (JP, A) JP-A-5-274261 (JP, A) JP-A-1-206451 (JP, A) JP-A-5-35701 (JP , A) JP 5-151177 (JP, A) JP 2-148248 (JP, A) JP 3-278238 (JP, A) JP 54-18252 (JP, A) JP 4-52953 (JP, A) JP-A-2-238556 (JP, A) JP-A-60-173636 (JP, A) JP-A-6-139209 (JP, A) JP-A-6-131282 (JP, A) JP 5-151179 (JP, A) JP 5-134959 (JP, A) JP 5-12228 (JP, A) JP 5-282261 (JP, A) JP 4 -267465 (JP, A) JP 63-211060 (JP, A) JP 5-12164 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) G06F 15/16- 15/177

Claims (4)

(57) [Claims] 1. A plurality of processing systems that transfer jobs as necessary through mutual communication , and when at least one of the processing systems fails, hot standby operation by other processing systems is possible. Information processing system configured in
Oite the beam, for each of the various jobs, a system distributing te <br/> over b le for sorting in accordance with the processing system to be run at system operating conditions is previously provided, each processing system, the system to monitor the operating state, when a job is given, referring to the system allocation table in light of the system operation status at that time, with respect to given the obtained job, during normal operation, the system distribution table Distribute the processing system of its own system or other system
When hot standby operation is performed by executing the first system allocation process
The above-mentioned No. 1 that was being executed by another processing system in which a failure occurred.
The second system allocation process as an alternative process to the first system allocation process
To execute the second system allocation process and the first
Job transfer between the system distribution processing of
A load distribution method characterized in that the load distribution method is performed together with the first system allocation processing . 2. When the system operating state is a low load state, the processing system to be executed is fixedly assigned to each of the various jobs, and when the system operating state is a high load state, the various job types are executed. The load balancing method according to claim 1, wherein a processing system to be executed is dynamically assigned to each of the above. Wherein each processing system, for the job received from another processing system, said system distributing claim, characterized in that running in the autologous without reference to the tables 1 or claim 2 The described load balancing method. 4. Jobs are received as necessary through mutual communication.
It has multiple processing systems for handing over, and at least one
When a failure occurs in a processing system, the
Information processing system configured to enable standby operation
And the processing system to be executed for each of the various jobs.
A system distribution table for distributing according to the system operating status
A table is provided, and each processing system has means for monitoring the system operation status and the system operation status at the time when a job is given.
The means for referring to the system allocation table based on the status of the system, and the system allocation table for a given job during normal operation.
Distributing the own processing system or other processing system indicated by the table
1 means for executing system allocation processing and another processing system in which a failure has occurred during hot standby operation
The other processing system executes the job given to
The second system as an alternative process of the first system allocation process
The distribution process is executed, and the second system distribution process and self
Jobs are delivered between the first system distribution process of
Processing is performed together with the above-mentioned first system distribution processing of itself.
An information processing system characterized by comprising means and .