JPH0934730A - Distributed processing method and distributed processor for the same - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the high-speed property and reliability of a system in the case of distributedly processing a processing request by taking charge of only a queue, to which the processing request adding a specified identifier is distributed, for at least one of main processing bodies. SOLUTION: The identifier concerning the length of execution time or priority is added to a processing request 1. A processing reception part 2 distinguishes the processing requests sent from clients, etc., into the request to be executed by itself and the request to be moved to the next queuing based on their identifiers and when executing the request by itself, the executed result is reported to the processing request source. The processing request not to be the executing object is queued by respective queues 4-6 corresponding to the identifiers at a queuing part 3 and main processing bodies 7-9 execute these requests but the main processing body 7 extracts only the processing request of the queue 4 corresponding to a most preferential identifier A. Therefore, even while the main processing bodies 8 and 9 just execute the arbitrary processing requests, the processing request of the identifier A is immediately executed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a distributed processing method and a distributed processing apparatus for the same, and more particularly, to a processing accepting unit which receives a processing request with an identifier relating to the length of execution time, priority, etc. The present invention relates to distributed processing in which a processing entity such as a process or a CPU individually executes the processing requests extracted from these queue groups and then individually distributes the processing requests.

In a distributed processing system in which a processing request sent from a server or the like is executed by a plurality of processes or CPUs, the individuality of each processing request, for example, the length of the execution time or the execution priority of the processing request itself is taken into consideration. It is desirable to perform the above-mentioned dispersion processing, and the present invention meets such a demand.

[0003]

2. Description of the Related Art FIG. 7 is an explanatory diagram showing an outline of a conventional distributed processing system. Reference numerals 51 to 53 relate to processing requesting devices such as clients and exchanges, and reference numerals 54 to 56 relate to long and short execution times and priorities. A processing request with an identifier (A to C), 57 is a server, 58 is a processing reception unit, 59 is a main queue, 60 to 62 are individual queues, 63 is a queue management unit that manages the main queue and individual queues, and 64 to 66 is a process as a processing body, 67 is a system definition file holding the correspondence between the identifier of the processing request and the individual queue, 68 is an exchange, 69 is a modem,
Reference numerals 70 denote responses to the processing request source. In the present specification, for convenience of explanation, the priority of processing is low in the order of identifier A-identifier B-identifier C -...

Here, for example, the client 52 is the identifier B.
When the processing request 55 having the above is issued, the processing request 55 is first sent to the processing accepting unit 58, then queued in the main queue 59, and then queued in the individual queue 61 corresponding to the identifier B.

When queuing from the main queue 59 to the individual queue 61, the queue management unit 63 confirms that the individual queue 61 corresponds to the identifier B by referring to the system definition file 67.

Among the processes 64-66, those waiting for a processing request check the contents of the individual queues 60-62 in the order of "identifier A-identifier B-identifier C -..." One of the processing requests is executed and the result is notified to the processing request source client 52 in the form of a response 70.

The processing requests 54 to 56 sent from the respective processing devices 51 to 53 to the server 57 are added with an identifier as well as a processing request device ID indicating the issuing source thereof, and each process 64.
~ 66 will recognize the destination of the execution result based on this ID.

FIG. 8 is an explanatory view showing a conventional queue management system. (A) shows that each of the processes 64 to 66 is waiting for a processing request and immediately sends the next processing request to the individual queues 60 to 62. (B) indicates that processes 64 and 65 are executing the processing request for the identifier B, and the process 64 is executing the processing request for the identifier C immediately after the next processing. It shows a state in which the highest priority processing request (identifier A) cannot be executed.

In the case of FIG. 8A, first, two processing requests in the highest priority individual queue 60 are, for example, processes.
One of the processing requests in the individual queue 61 is captured by the process 66.

In the case of FIG. 8B, each process 64-66 cannot fetch the next processing request from the individual queue until the execution of the current processing request is completed, and therefore, the individual queue having the highest priority is processed. The processing request contained in 60 will not be taken into there until the processing in any process is completed.

[0011]

The conventional distributed processing method as described above is as follows: When a processing request whose execution time is short, for example, a processing request such as login or logoff, is sent from the processing accepting unit to the main queue and the individual queue. Since it is queued, the response time to a processing request becomes long, and even if a high-priority processing request is queued in an individual queue, if all processes are running at that time, the processing request will be processed individually. Since there is a waiting state (overload state) that remains in the queue, there is a problem that processing requests with high priority are postponed and inadequacy remains in terms of system speed and reliability. are doing.

Therefore, according to the present invention: When the processing accepting unit determines from the identifier of a processing request that the execution time is short, the processing request is executed by itself without queuing, and the result is sent to the processing request source. The execution time by ensuring that at least one of the process and the CPU that is the processing subject of the processing request is in charge of only the processing request having a specific identifier (for example, the identifier with the highest priority). It is an object of the present invention to further improve the speed and reliability of the system when performing distributed processing of processing requests that differ in terms of priority.

[0013]

FIG. 1 is a diagram for explaining the principle of the present invention. In the figure, reference numeral 1 denotes a processing request, which has a form in which an identifier relating to the length of execution time or priority thereof is added. Reference numeral 2 denotes a processing reception unit, which processes a processing request sent from a client or the like based on its identifier.
When the processing request belongs to the former, the execution result is notified to the processing request source when the processing request itself is distinguished from the one to be transferred to the next queuing. A queuing unit 3 distributes the processing request sent from the processing reception unit 2 to a queue corresponding to the identifier. Reference numeral 4 denotes a queue (No. 1), which corresponds to the identifier A. 5
Is a queue (No. 2) and corresponds to the identifier B. Reference numeral 6 denotes a queue (No. 3), which corresponds to the identifier C. Reference numeral 7 denotes a processing entity (No. 1) such as a CPU and a process, and is in charge of only the queue 4 corresponding to the identifier A. Reference numeral 8 denotes a processing entity (No. 2) such as a CPU and a process, and is in charge of only the queue 4 corresponding to the identifier A and the queue 5 corresponding to the identifier B. Reference numeral 9 is a processing entity (part 3) such as a CPU and a process, and is in charge of all queues. Reference numeral 10 is a first response, which is notified from the processing accepting unit 2 to the processing request source. Reference numeral 11 is a second response, which is notified to the processing request source from any one of the processing subjects 7 to 9.

The process accepting unit 2 specifies the execution destination of the process request sent from the client or the like by the difference in the identifiers (for example, the numbers and alphabetical characters shown in FIG. 4), and login, logoff, etc. The processing request having a short execution time is executed by itself.

Then, the processing requests that have not been executed by the processing accepting unit 2 are queued in the queues 4 to 6 corresponding to the identifiers in the same manner as in the prior art, and the processing subjects 7 to 9 process these. Although it is executed after taking out, in the present invention, the relationship between each of the processing subjects and the queues that can be taken out therefrom is specified by a process management table (see FIG. 5) described later.

This specific content indicates that at least one of the processing entities 7 to 9 is in charge of only the queue corresponding to the specific identifier. For example, the processing entity 7 is the highest priority identifier. It indicates that only the processing request of the queue 4 corresponding to A is taken out.

Therefore, the processing entity 7 does not execute a request having a lower priority than the processing request of the identifier A, and even if the processing entities 8 and 9 are executing the processing request of an arbitrary identifier, the queue 4 is processed. The processing request of the identifier A queued in is immediately executed by the processing entity 7.

However, when the processing subject 7 is executing the processing request for the identifier A and the other processing subjects 8 and 9 are also executing the processing request for the arbitrary identifiers, one of the processing subjects You have to wait for the end of execution.

In the above description, the identifier assigned to the processing request and the queue holding the processing request have a one-to-one correspondence, but one queue may be associated with a plurality of identifiers. Good.

The basic configuration (1) as the distributed processing method of the present invention is that "a processing request to which an identifier relating to the length of execution time, priority, etc. is assigned is once distributed to a queue corresponding to the identifier, and then, In a distributed processing method in which a processing entity such as a process or a CPU extracts the processing request from these queue groups and individually executes the processing request, at least one of the processing entities is assigned the processing request with the specific identifier. I was in charge of only the queue. "

The basic configuration (No. 2) of the distributed processing method of the present invention corresponds to "the processing receiving unit receives a processing request with an identifier relating to the length of execution time, priority, etc., and then corresponds to the identifier. In a distributed processing method in which a processing entity such as a process or a CPU takes out the processing request from these queue groups and executes the processing requests individually, the processing receiving unit determines the execution time in the processing request. Those having the identifier indicating that it is short are executed by themselves, and the result is notified to the processing request source. "

The basic configuration (No. 1) of the distributed processing apparatus of the present invention is that "a processing request to which an identifier relating to the length of execution time, priority, etc. is assigned is once distributed to a queue corresponding to the identifier, and then, In a distributed processing device in which a processing entity such as a process or a CPU extracts the processing request from these queue groups and individually executes the processing request, at least one of the processing entities is assigned the processing request with the specific identifier. It has a function of executing only the contents of the queue. "

The basic configuration (No. 2) of the distributed processing apparatus of the present invention is "corresponding to the identifier after the processing request with an identifier relating to the length of execution time, priority, etc. is received by the processing accepting unit. In a distributed processing device in which a processing entity such as a process or a CPU extracts the processing request from these queue groups and individually executes the processing request, the processing accepting unit determines the execution time in the processing request. Those having the above-mentioned identifier indicating that they are short must have a function of executing them by themselves and notifying the result to the processing request source. ”

FIG. 2 is an explanatory view showing the queue management system of the present invention. (A) shows that each of the processing entities 7 to 9 is waiting for a processing request and immediately sends the next processing request to the queue 4 to 4. (B) is a state in which the processing subject 7 is waiting for a processing request, and the processing subjects 8 and 9 are in the process of executing the processing request of the identifier B,
Are shown respectively.

In the case of FIG. 2A, the two processing requests stored in the queue 4 corresponding to the identifier A are fetched by the processing entities 7 and 8 and are entered in the queue 5 corresponding to the identifier B. One processing request that has been made is taken in by the processing subject 9.

In the case of FIG. 2B, one processing request in the queue 4 corresponding to the identifier A is fetched by the processing subject 7 dedicated to the identifier A without waiting, but corresponds to the identifier C. One processing request in the queue 6 is not taken into the processing body 8 or 9 until the processing body 8 or 9 finishes executing the current processing request.

[0027]

DETAILED DESCRIPTION OF THE INVENTION An embodiment of the present invention will be described with reference to FIGS. In the following description, for convenience of description, it is assumed that a process is used as a processing subject.

FIG. 3 is an explanatory view showing an outline of the distributed processing system of the present invention. 21 to 23 are processing requesting devices such as clients, and 24 to 26 are identifiers relating to the length of execution time and priority. Processing request, 27 is a server, 28 is a process receiving unit, 29 is a main queue, 30 to 32 are individual queues, 33 is a queue management unit that manages the main queue and individual queues, and 34 to 36.
Is a process, 37 is a process management table (see FIG. 5),
38 is an exchange, 39 is a modem, 40 is a system definition file,
Reference numeral 41 denotes the first response from the process reception unit 28, and 42 denotes the second response from the processes 34 to 36.

The system definition file 40 includes: -element data of the identifier table (see Fig. 4) -correspondence data of identifiers and individual queues (see Fig. 7) -element data of the process management table 37 (see Fig. 5) Are stored, and these data can be changed by the user. The server 27 may set the data independently without preparing these data in the system definition file 40.

The distributed processing method of the present invention and that of the conventional method (see FIG. 7).
The main difference is that the process receiving unit 28 that has received a process request from an arbitrary client or the like executes the process request by itself when the identifier is a numerical value, and Response to 41
The process 34 is dedicated to the individual queue 30 in which the processing request of the identifier A corresponding to the highest priority is placed, that is, the process 34 processes the individual queues 31 and 32. Do not take out requests ・ Providing a process management table 37 that indicates which individual queue processing requests can be taken out by each process.

FIG. 4 is an explanatory diagram showing the identifier table of the present invention. In this identifier table, the identifiers and the processing contents are associated with each processing request, and numbers are set for the identifiers of processing requests such as login and logoff that require a short processing time. The "update table" and "pager center" that appear in the description of the processing contents are omitted in FIG.

FIG. 5 is an explanatory diagram showing the process management table of the present invention. The process management table 37 holds: an individual queue that can be taken out; a processing capacity; a current processing status (whether or not processing is being performed);

In the illustrated process management table 37, the process 34 can take out only the processing request of the individual queue 30, and the processing request of "incoming call" is "12:04 10.95.
It is started at "second" and is still running.-Process 35 can retrieve the processing requests of both individual queue 30 and individual queue 31, and is not currently executing the processing request.-Process For 36, the processing requests of all individual queues 30 to 32 can be taken out, and the processing request of "pager call" was started at "12: 04: 05.55 seconds" and is still being executed. Is held.

FIG. 6 is an explanatory diagram showing a processing procedure for a processing request according to the present invention, and the contents thereof are as follows. (1) The process receiving unit 28 receives the process request with the client ID and the like attached, and proceeds to the next step. (2) The process receiving unit 28 determines whether the identifier of the process request is a numerical value or an alphabetic character, and proceeds to step (10) in the former case and proceeds to the next step in the latter case. . (3) The queue management unit 33 sends the processing request to the main queue 29.
Queue and proceed to the next step. (4) The queue management unit 33 confirms the individual queue corresponding to the identifier of the processing request by referring to the system definition file 40 as in the case of FIG. 7, and proceeds to the next step. (5) The queue management unit 33 queues the processing request in the individual queue and proceeds to the next step. (6) Each process waiting for a processing request confirms the accessible individual queue by referring to the process management table 37 of FIG. 5, and proceeds to the next step. (7) The process executes the processing request fetched from the individual queue and proceeds to the next step. (8) The process sends the execution result to the second response.
As 42, the processing requesting client is notified, and the process proceeds to the next step. (9) The queue management unit 33 determines whether or not all the processing requests in the individual queues 30 to 32 have been taken in by each process, and if “YES”, terminates the series of processing,
If "NO", return to step (6). (10) The process receiving unit 28 executes the process request and proceeds to the next step. (11) The process reception unit 28 uses the execution result as the first response.
As 41, the client of the processing request source is notified and the series of processing is ended.

In step (6), each process waiting for a processing request checks each accessible individual queue in descending order of priority, and confirms whether or not there is a processing request.

When there are a plurality of processes that can be the execution subject of a processing request in a queue with a higher priority, it is desirable to execute the processing request by the process with the higher processing capacity.

[0037]

As described above, according to the present invention, when distributed processing is performed for each processing request having a difference in execution time or priority, the execution time is shortened from the processing request identifier in the processing receiving unit. When it is determined that the processing request is executed by itself without queuing, the result is notified to the processing request source, and a specific identifier ( For example, assign it to the execution of only processing requests with the highest priority identifier),
Therefore, the high speed and reliability of the distributed processing system can be further improved.

[Brief description of drawings]

FIG. 1 is a diagram illustrating the principle of the present invention.

FIG. 2 is an explanatory diagram showing a queue management system of the present invention.

FIG. 3 is an explanatory diagram showing an outline of a distributed processing system of the present invention.

FIG. 4 is an explanatory diagram showing an identifier table of the present invention.

FIG. 5 is an explanatory diagram showing a process management table according to the present invention.

FIG. 6 is an explanatory diagram showing a processing procedure for a processing request according to the present invention.

FIG. 7 is an explanatory diagram showing an outline of a conventional processing distribution system.

FIG. 8 is an explanatory diagram showing a conventional queue management system.

[Explanation of symbols]

In FIG. 1, 1 ... Process request to which an identifier relating to the length of execution time or priority is attached 2 ... Process accepting unit 3 ... Queuing unit 4 ... Queue (identifier A: highest priority) 5・ ・ ・ Queue (identifier B) 6 ... Queue (identifier C) 7 ... Processing subject (only responsible for queue 4) 8 ... Processing subject (only responsible for queues 4 and 5) 9 ... Processing Principal (responsible for all queues) 10 ... First response 11 ... Second response

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Junyo Taoka 1015 Kamiodanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture, Fujitsu Limited (72) Inventor Masahiko Murakami, 1015, Kamikodanaka, Nakahara-ku, Kawasaki, Kanagawa Prefecture, Fujitsu Limited ( 72) Inventor Tadashige Iwao 1015 Kamiodanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture, Fujitsu Limited (72) Inventor Fukuyama, Noriyuki Fukuyama, Kanagawa Prefecture 1015, Kamedotachu, Nakahara-ku, Kawasaki, Kanagawa Prefecture (72) Inventor, Masanobu Morinaga Kanagawa Prefecture 1015 Kamiodanaka, Nakahara-ku, Kawasaki City, Fujitsu Limited

Claims (8)

[Claims] 1. A processing request to which an identifier relating to the length of execution time, priority, etc. is assigned to a queue corresponding to the identifier, and then a processing entity such as a process or a CPU sends the processing request from these queue groups. In the distributed processing method of taking out and individually executing, at least one of the processing main bodies is in charge of only the queue to which the processing request with the specific identifier is distributed. . 2. A processing request, to which a processing request to which an identifier relating to the length of execution time, priority, etc. is attached, is received, and then distributed to a queue corresponding to the identifier, and then a processing entity such as a process or a CPU processes the processing. In the distributed processing method of fetching requests from these queue groups and individually executing the requests, the process receiving unit executes itself for those having the identifier indicating that the execution time is short among the processing requests, A distributed processing method characterized in that the result is notified to a processing request source. 3. The distributed processing method according to claim 2, wherein at least one of the processing entities is in charge of only the queue to which the processing request with the specific identifier is distributed. 4. The distributed processing method according to claim 1, wherein the specific identifier indicates a highest priority for execution of the processing request. 5. A processing request to which an identifier relating to the length of execution time, priority, etc. is assigned to a queue corresponding to the identifier, and then a processing entity such as a process or a CPU sends the processing request from these queue groups. In the distributed processing device for taking out and individually executing, at least one of the processing main bodies has a function of executing only the contents of the queue to which the processing request with the specific identifier is distributed. Characterized distributed processing device. 6. A processing request, to which a processing request to which an identifier relating to the length of execution time, priority, etc. has been added, is received by a processing accepting unit, and once distributed to a queue corresponding to the identifier, and then a processing entity such as a process or a CPU processes the processing. In a distributed processing device that retrieves requests from these queue groups and executes them individually, the process accepting unit executes itself for those with the identifier indicating that the execution time is short in the process requests, A distributed processing device having a function of notifying a processing request source of the result. 7. The at least one of the processing subjects has a function of executing only the contents of the queue to which the processing request to which the specific identifier is attached is distributed. The distributed processing device described. 8. The distributed processing device according to claim 5, wherein the specific identifier indicates the highest priority for execution of the processing request.