JPH07319834A - Load distribution system in distributed on-line system - Google Patents

Abstract

PURPOSE:To efficiently distribute processing requests as loads and to make system operation flexible in a distributed on-line system connecting plural computers through a network and providing the same service from plural computers thereby executing load distribution. CONSTITUTION:In the distributed on-line system connecting plural computers through the network, respective request scheduling computers 11, 12 are provided with queues 31, 32 for receiving request processing, schedule processors 41 to 46 for scheduling processing requests from the queues 31, 32 and a schedule process management table 51, 52 and request processing computers 13 to 15 are provided with request processing processes 61 to 66.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a distributed online system that provides the same service with a plurality of computers, and more particularly to a load distribution system for the distributed online system.

[0002]

2. Description of the Related Art In a conventional online system, when load distribution is performed among a plurality of computers, the response time to a processing request is measured and the processing request is issued to the place where the response time is shortest. The load balancing method between multiple computers in the online system is described on page 168 of "To Build a Business System in UNIX Environment-A Complete Guide to RDBMS and TP Monitors" (Nikkei BP, published July 30, 1993). Has been.

[0003]

The above-mentioned prior art does not consider variations in the processing time of individual computers, and if the processing time for a certain processing request takes a long time, the processing may be performed by another computer. However, there is a problem that other requests already scheduled to the same computer are not processed and become inefficient. 1 of the present invention
One purpose is to balance the load among multiple computers.
There is a more efficient load distribution.

Further, in the above-mentioned conventional technique, there is no means for cutting off a request from the outside when a part of the request processing computer is removed from the online work and the batch work is desired.
One object of the present invention is to enable a flexible system operation, such as removing a part of the request processing computer from online business and performing batch business.

[0005]

The first object of the present invention is to:
When the load is distributed among a plurality of computers, it can be achieved by separating into a request scheduling computer for queuing and scheduling processing requests and a request processing computer for processing the requests.

The second object of the present invention is achieved by providing a command for stopping and restarting the function of the above schedule process.

[0007]

According to the present invention, when load distribution is performed among a plurality of computers, a request schedule computer for queuing and scheduling processing requests and a request processing computer for processing requests are separated and scheduled. A process schedules a processing request for a request processing process. As a result, the queued processing request can be efficiently processed. It also has commands to stop and restart the scheduling of the scheduling process. As a result, flexible system operation is possible, such as stopping the scheduling of some of the computers performing online processing and performing batch jobs.

[0008]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a configuration example of a distributed processing system to which the present invention is applied. In this embodiment, the computer 11,
12, 13, 14, and 15 are connected to the network 10. The computers 11 and 12 are computers for request schedules, and communication lines 21 and 22 connected to the outside.
There are queues 31 and 32 for receiving processing requests from the communication lines, schedule processes 41 to 46 for scheduling processing requests from the queues, and schedule process management tables 51 and 52. Calculator 1
Request processing computers 3, 14 and 15 are provided with request processing processes 61 to 66. In FIG. 1, processing requests queued in the queue 31 from the communication line 21 are scheduled by the schedule processes 41, 42, and 43 to the request processing processes 61, 63, and 65, respectively.
The processing requests queued in the queue 32 from the communication line 22 are shown to be scheduled by the schedule processes 44, 45, 46 to the request processing processes 62, 64, 66, respectively.

FIG. 2 is a diagram showing a configuration example of the schedule process management table in FIG. Process I of the schedule process to support the schedule process
D71 and computer name 72 for scheduling the processing request
The process request transmission destination process address 73, the schedule status 74, and the queue ID of the queue in which the process request is queued. The schedule state is rewritten by a command and is set to "0" when the schedule process is stopped, "1" when the schedule process is scheduled, and "2" when a process stop request is issued.

FIG. 3 is a flowchart showing the processing request transmission processing of the schedule process in FIG. In this embodiment, if the schedule state of the schedule process is "2" (1001), the process is immediately terminated. If the schedule state is "0" (1002), wait for 1 second, and (1004) process is restarted from the beginning.
If the schedule status is "1" (1002) and there is a processing request in the queue (1003), the processing request is extracted from the queue (1005), and the processing request is sent to the request processing process indicated by the destination address of the schedule process management table. Is transmitted (1006), the processing result of the request processing process is received (1007), and the processing result is transmitted to the processing request source (1008). If there is no processing request in the queue (1003), wait for 1 second (1004), and then perform processing again from the beginning.

FIG. 4 is a flow chart showing the operation of the request processing of the request processing process in FIG. In this embodiment, when a processing request is received (2001), the request is processed (2002) and the processing result is transmitted (2003).

FIG. 5 is a flow chart showing the operation of the command for stopping the schedule of the schedule process in FIG. In this embodiment, the computer name for stopping the schedule and the queue ID are specified in the command. The command searches for an entry in the schedule process management table that has the specified computer name and queue ID (3
001) and set the schedule status to "0" (300
2). Note that the present invention is not limited to the system configuration shown in FIG. 1, and the number of computers, processes,
It is applicable to a system including a queue.

[0013]

According to the present invention, it is possible to efficiently distribute the load in a distributed online system that distributes the load among a plurality of computers. It is also possible to give flexibility to system operation.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration example of a distributed processing system to which the present invention is applied.

FIG. 2 is a diagram showing a configuration example of a process request schedule process management table in FIG.

FIG. 3 is a flowchart showing a process request transmission process of the schedule process in FIG.

FIG. 4 is a flowchart showing the operation of the request processing process in FIG.

5 is a flowchart showing an operation of a command for stopping the schedule of the schedule process in FIG.

[Explanation of symbols]

 10 ... Communication network, 11 ... Request schedule computer, 12 ... Request schedule computer, 13 ... Request processing computer, 14 ... Request processing computer, 15 ... Request processing computer, 21 ... Communication line, 22 ... Communication line, 31 ... Queue, 32 ... Queue, 41 ... Schedule process, 42 ... Schedule process, 43 ... Schedule process, 44 ... Schedule process, 45 ... Schedule process, 46 ... Schedule process, 51 ... Schedule process management table, 52 ... Schedule process management Table, 61 ... Request processing process, 62 ... Request processing process, 63 ... Request processing process, 64 ... Request processing process, 65 ... Request processing process, 66 ... Request processing process.

Claims (2)

[Claims] 1. A distributed online system in which a plurality of computers are connected via a network, the same service is provided by the plurality of computers, and load distribution is performed, and one or a plurality of request schedules for accepting data processing requests. Computer, one or more request processing computers for processing the received requests, a queue for queuing processing requests in the request schedule computer, and the request in the request schedule computer Provide a schedule process corresponding to the processing computer and a management table of the schedule process, have a request processing process corresponding to the schedule process in the request processing computer, and the schedule process sends a processing request to the request processing process. Distributed online that features scheduling Load-balancing method of the system. 2. A load balancing method for a distributed online system, wherein in addition to the load balancing method according to claim 1, the function of the schedule process can be stopped and restarted by a command.