JP2004220083A - Message queueing method and message queueing system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To change the priority of a request temporarily retaining in an input/output queue according to a change in the state of a computer at the request origin. <P>SOLUTION: In a computer 10, when a request is extracted from an input/output queue, the priority of the requests is determined by referring to not only priority information added during the request but also the priority information managed by a request origin priority managing part 11. In the request origin priority managing part 11, the priority information is managed by directly receiving the value of the priority information from priority management parts 22 and 23 of computers 20 and 30 through a network. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a technique for controlling a message queue, and in particular, in a distributed processing environment in which a program is executed while exchanging data between a plurality of computers via a network, requests and responses from other computers can be efficiently performed. The present invention relates to a message queue management technique suitable for managing a message queue.
[0002]
[Prior art]
Conventionally, a technique for controlling writing to a disk using a message queue is described in Patent Literature 1, for example. In addition, the presence or absence of occurrence of transmission abnormality between tasks in a computer device or between computer devices using a message queue A technique for detecting a situation and a situation is described in Patent Document 2, for example.
[0003]
Further, in a distributed processing environment including a plurality of computers, management of a processing order based on a message queue is performed. Hereinafter, a conventional priority management technique for exchanging a request from another computer and a response to the other computer via an input / output queue will be described with reference to FIG.
[0004]
FIG. 6 is a block diagram showing a configuration example of a distributed processing system using a conventional message queue management technique.
[0005]
In FIG. 6, computers 10a, 20a, and 30a are computers, and the computers 10a, 20a, and 30a are connected via networks 40a, and execute programs while exchanging data via the network 40a. We are building a distributed processing environment.
[0006]
The client programs 21a and 31a in the computers 20a and 30a are programs that make requests to other computers 10a. The server program 12a of the computer 10a receives and processes the requests from the client programs 21a and 31a, and responds. Is a program that returns
[0007]
In such a configuration, in the distributed processing system, for example, a request from the client program 21a on the computer 20a is transmitted to the computer 10a via the network 40a.
[0008]
In the computer 10a, the request from the client program 21a is passed directly to the server program 12a, or is passed to the server program 12a via the input queue management unit 13a. In addition, a response as a result of processing the request by the server program 12a is directly returned to the client program 21a or returned via the output queue management unit 14a.
[0009]
When exchanging via a queue, each request can be registered in the queue with a priority. When requests from a plurality of client programs (21a, 31a) arrive at the computer 10a at the same time and stay in the input queue management unit 13a, requests with higher priority are processed before requests with lower priority. It has become.
[0010]
That is, in the input queue management unit 13a, when a request is taken out of the queue, the priority added to each request is checked, and the request with the highest priority is taken out of the queue first.
[0011]
Such a technique is also employed in the UNIX (registered trademark) message queue system call, and is described in, for example, Non-Patent Document 1 and the like.
[0012]
However, according to this conventional technique, the priority of the message queue is set before the request is put into the queue, and after the request temporarily stays in the queue, the priority is changed even if the state of the request source changes. I can't.
[0013]
That is, even if the state suddenly changes immediately after the request of the client program 21a stays in the input queue management unit 13a and the urgency is no longer required, the priority of the request staying in the input queue management unit 13a is determined. It cannot be lowered.
[0014]
For this reason, there has been a problem that requests that no longer require urgency are processed first on the server.
[0015]
[Patent Document 1]
JP 2002-245714 A [Patent Document 2]
JP 2002-189605 A [Non-Patent Document 1]
W. Richard Stevens, translated by Yoichi Shinoda, "UNIX (R) Network Programming 3.9 Message Queue" (published by Toppan Co., Ltd.)
[0016]
[Problems to be solved by the invention]
The problem to be solved is that, in the prior art, once a request has stayed in the input / output queue, the priority cannot be changed even if the state of the requesting computer changes.
[0017]
An object of the present invention is to solve these problems of the prior art. For example, when the state of the requesting side changes and the urgency is no longer necessary, the request staying in the input / output queue is By dynamically lowering the priority, it is possible to improve the response time of other requests and to efficiently process the requests in the entire system.
[0018]
[Means for Solving the Problems]
In order to achieve the above object, according to the present invention, when a request is taken out from an input / output queue (16) in a computer (10), not only the conventional priority information added in the request but also the newly added request The priority is also determined with reference to the priority information (priority change information) managed by the original priority management unit (11). The request source priority management unit (11) directly receives and manages information for changing the priority sent from the request source computer (20, 30) via the network (40).
[0019]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
[0020]
FIG. 1 is a block diagram showing a configuration example of a distributed processing device provided with a message queue management system according to the present invention, and FIG. 2 is an explanatory diagram showing a first operation example of the message queue management system according to the present invention. , FIG. 3 is an explanatory diagram showing a second operation example of the message queue management system according to the present invention, FIG. 4 is an explanatory diagram showing a third operation example of the message queue management system according to the present invention, and FIG. FIG. 14 is an explanatory diagram showing a fourth operation example of the message queue management system according to the present invention.
[0021]
The message queue management system according to the present invention is realized by a computer configuration including a CPU (Central Processing Unit), a main memory, a display device, an input device, and an external storage device. After the programs and data recorded on the optical disk are installed in the external storage device, the programs and data are read from the external storage device into the main memory and processed by the CPU, so that the respective processing units in the computers 10, 20, and 30 shown in FIG. Is constructed.
[0022]
In FIG. 1, a distributed processing environment is configured by a computer 10 and computers 20 and 30 connected via a network 40. The computer 10 includes a server program 12, an input queue management unit 13, an output queue management unit 14, An original priority management unit 11 is provided.
[0023]
The computer 20 includes a client program 21 and a priority management unit 22. Similarly, the computer 30 includes a client program 31 and a priority management unit 32.
[0024]
As described above, in the message queue management system of this example, the request source priority management unit 11 in the computer 10 and the priority management units 22 and 22 in the computers 20 and 30 are different from the conventional message queue management system shown in FIG. 32 are provided.
[0025]
Hereinafter, how the requests from the client programs 21 and 31 of the computers 20 and 30 are processed by the input queue management unit 13 in the computer 10 will be described with reference to FIG.
[0026]
In FIG. 2, as shown by 50 to 52, the requests a to c of the computers 20 and 30 that are staying and managed in the input queue management unit 13 of the computer 10 include priority information (priority: 1, 3) and computer identification information (computers: computers 20, 30).
[0027]
The priority information (priority: 1, 3) indicates the priority of the request. Here, an integer value from “1” to “5” is used as a priority as an example. Further, as for the numerical value of the priority, a smaller numerical value indicates a higher priority than a larger numerical value.
[0028]
The computer identification information (computers: computers 20 and 30) is information for specifying the requesting computer. The computer identification information (computers: computers 20 and 30) is used when the server program 12 of the computer 10 determines which computer returns a response after processing the request.
[0029]
The request source priority management unit 11 of the computer 10 includes a request source priority management table 15, and uses this request source priority management table 15 to determine the priority for each request source computer (20, 30). Managing.
[0030]
The number of priorities managed in the request source priority management table 15 may be one for one computer (20, 30), that is, the priority may be managed for each computer 20, 30. A plurality may exist for each request of each computer (20, 30). In the example shown in FIG. 2, one priority is set for one computer (20, 30).
[0031]
The priority value managed in the request source priority management table 15 may be a relative value to the priority information added in the request, or may be a fixed value for each computer (20, 30). It is good. In the example shown in FIG. 2, a relative value with respect to the priority information added in the request is set, and the relative priority with respect to the requests from the computers 20 and 30 is ± 0.
[0032]
In the input queue management unit 13 of the computer 10, three requests (request a, request b, and request c) are in the input queue 16. Further, it shows that the requests arrive at the input queue 16 in the order of (1) request a (50), (2) request b (51), and (3) request c (52). .
[0033]
The request a (50) is a request from the computer 20 and has a value of priority 3. The request b (51) is a request from the computer 30 and has a value of priority 1. The request c (52) is a request from the computer 20 and has a value of priority 3. It is in a state of waiting for these three requests a50 to c52 to be processed by the server program 12.
[0034]
When taking out the requests (a50 to c52) from the input queue 16 and passing them to the server program 12, the input queue management unit 13 first stores the priority information of all the requests (a50 to c52) registered in the input queue 16 inspect.
[0035]
At this time, the input queue management unit 13 adds the priority value corresponding to the request source computer registered in the request source priority management table 15 to the value of the priority information in the requests (a50 to c52). Is used as the final priority value.
[0036]
The one with the lowest final priority value, that is, the one with the highest priority is taken out of the queue first. In the case of the same value, the request registered earlier has priority. According to this algorithm, the three requests (a50 to c52) shown in FIG. 2 are taken out of the input queue 16 in the order of the request b51, the request a50, and the request c52.
[0037]
Next, a management control technique of the request source priority management unit 11 in the computer 10 will be described with reference to FIG. FIG. 3 shows the relationship between the request source priority management unit 11 in the server computer 10 and the priority management units 22 and 32 in the client computers 20 and 30 respectively.
[0038]
Here, a case where it is desired to lower the priority for a request from the computer 30 is considered. In the computer 30, the priority management is performed by the priority management unit 32. The computer 30 changes the priority value in the priority management table 33 from “± 0” to “+3” based on an instruction from the operator or the like, and then changes this value via the network 40 to the computer 10. To the request source priority management unit 11.
[0039]
The request source priority management unit 11 of the computer 10 that has received the notification changes and sets the priority value in the column corresponding to the computer 30 in the request source priority management table 15 from “± 0” to “+3”. As a result, the request from the computer 30 is processed with its priority lowered by three levels when it is taken out of the input queue 16.
[0040]
FIG. 4 shows the management state of the computer 10 after the priority of the computer 30 has been changed from “± 0” to “+3”.
[0041]
As shown in FIG. 4, in this case, the order of the requests a50 to c52 that will be executed on the server program 12 is (1) request a50, (2) request c52, and (3) request b51.
[0042]
As described above, the priority of the request b51 having the highest priority at the time of staying in the input queue 16 in the input queue management unit 13 of the computer 10 shown in FIG. 2 is determined by the priority management 32 in the computer 30 shown in FIG. By changing the information of the request source priority management unit table 15 based on the operation in the above, it can be seen that the value has been changed to the lowest value.
[0043]
The change timing of the request source priority management unit table 15 is not changed by receiving the notification from the priority management units 22 and 32 at the time of the change on each of the computers 20 and 30, but by the request of the computer 10. The original priority management unit 11 may make an inquiry to the priority management units 22, 32,... On all the computers (20, 30,...) At regular time intervals.
[0044]
FIG. 5 shows that the request source priority management unit 11 of the computer 10 inquires the priority management units 22, 32,... On all the computers (20, 30,. By doing so, the operation when the request source priority management table 15 is changed is shown.
[0045]
For example, consider a case where a low priority is set for a request from the busy computer 20 and a high priority is set for a request from a vacant (no processing) computer 30.
[0046]
The request source priority management unit 11 inquires the priority management units 22 and 32 on the computers 20 and 30 at regular intervals, and the priority management units 22 and 32 respectively check the computers 20 and 30 at that time. Of the CPU usage rate and converts it into a priority.
[0047]
By responding this priority to the request source priority management unit 11 of the computer 10, new priority information is always set in the request source priority management management table 15 of the request source priority management unit 11. Thereby, the requests (a to c) staying in the input queue 13 can be taken out with the priority reflecting the latest state.
[0048]
Although the management / control processing operation for the input queue management unit 13 has been described above, the management / control of the output queue management unit 14 in FIG. 1 can be performed in exactly the same manner as the input queue management unit 13. .
[0049]
For example, when taking out a request from an output queue, the output queue management unit 14 refers to the priority information of the request source priority management unit 11 to dynamically determine the order in which responses are returned to the computers 20 and 30. It can be changed.
[0050]
As described above with reference to FIGS. 1 to 5, in the present example, the server-side computer 10 directly receives the priority information transmitted from each of the client-side computers 20 and 30 via the network 40. Then, when the request source priority management unit 11 manages the request in the request source priority management table 15 and takes out the request from the input / output queue, the input queue management unit 13 and the output queue management unit 14 request the request source priority management unit. 11 refers to the priority information managed in the request source priority management table 15 and newly obtains not only the conventional priority information added in each request (a50 to c52) but also a new request source priority management unit. The priority is also determined by referring to the 11 priority information.
[0051]
As a result, when a request from the client program 21 of the computer 20 stays in the input queue management unit 13 of the computer 10 on the server side, for example, the state of the computer 20 suddenly changes and urgency is no longer required. In this case, the priority of the request staying on the input queue management unit 13 can be lowered, and the problem that a request that no longer requires urgency is processed first on the server does not occur.
[0052]
The present invention is not limited to the examples described with reference to FIGS. 1 to 5 and can be variously modified without departing from the gist thereof. For example, in the present example, the distributed processing system configuration includes the server-side computer 10 and the client-side computers 20 and 30 connected via the network 40. However, a system having a configuration in which a plurality of CPUs are connected by a bus may be used. Applicable.
[0053]
Further, the computers 20 and 30 are used as Web server devices in each ISP (Internet Services Provider) connected via the Internet, and the computer 10 is used in response to requests from, for example, an application server program or a CGI (Common Gateway Interface) program of each Web server device. It may be configured to be a database that performs data processing accordingly.
[0054]
Further, as an example of a computer configuration for realizing the system of the present example, a computer configuration without a keyboard or an optical disk driving device may be used. In this example, the optical disk is used as the recording medium, but an FD (Flexible Disk) or the like may be used as the recording medium. As for the installation of the program, the program may be downloaded and installed via a network via a communication device.
[0055]
【The invention's effect】
According to the present invention, the priority of a request (queue) can be changed even after the request temporarily stays in the input / output queue on the computer. For example, the state of the computer that has requested the request changes. When urgency is no longer required, the response time of other requests can be improved by dynamically lowering the priority of the request staying in the input / output queue. As a result, it is possible to efficiently process requests in the entire system, and it is particularly effective in a system in which the processing time of the server program is long and the requests stay in the input / output queue for a long time.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a configuration example of a distributed processing device provided with a message queue management system according to the present invention.
FIG. 2 is an explanatory diagram showing a first operation example of the message queue management system according to the present invention.
FIG. 3 is an explanatory diagram showing a second operation example of the message queue management system according to the present invention.
FIG. 4 is an explanatory diagram showing a third operation example of the message queue management system according to the present invention.
FIG. 5 is an explanatory diagram showing a fourth operation example of the message queue management system according to the present invention.
FIG. 6 is a block diagram showing a configuration example of a distributed processing system using a conventional message queue management technique.
[Explanation of symbols]
10: computer (server), 11: request source priority management unit, 12: server program, 13: input queue management unit, 14: output queue management unit, 15: request source priority management table, 16: input queue, 20 , 30: computer (client), 21, 31: client program, 22, 32: priority management unit, 23, 33: priority management table, 40: network, 50: request a, 51: request b, 52: request c, 10a: computer (server), 12a: server program, 13a: input queue management unit, 14a: output queue management unit, 20a, 30a: computer (client), 21a, 31a: client program, 40a: network.

Claims (6)

A message queuing method for a message queuing system, which sequentially retrieves a plurality of requests staying in an input / output queue based on priority information added in advance to each request by a requester,
A first step of storing in a storage device priority change information notified from a request source of a request staying in the input / output queue;
A second step of specifying a request to be taken out preferentially based on the priority information previously added to each request and the priority change information stored in the storage device when taking out requests from the input / output queue A message queuing method comprising: extracting requests from an input / output queue in an order based on the priority information and the priority change information. A message queuing method of a message queuing system that sequentially retrieves each request from a plurality of computers on a network that is staying in an input / output queue based on priority information previously added to each request by a requesting computer. So,
A first step of storing, in a storage device, priority change information notified from a request source computer of the request staying in the input / output queue via a network;
A second step of specifying a request to be taken out preferentially based on the priority information previously added to each request and the priority change information stored in the storage device when taking out requests from the input / output queue And having
A message queuing method, wherein requests are taken out of an input / output queue in an order based on the priority information and the priority change information. A message queuing method according to any one of claims 1 and 2,
The priority information and the priority change information include numerical values representing priorities,
In the first step, a numerical value of the priority change information is stored in the storage device for each request source,
In the second step, priority is given based on a result obtained by adding and subtracting a numerical value of the priority change information stored in the storage device in units of the request source and a numerical value of the priority information added in advance to each request. Message queuing method, wherein a request to be retrieved is specified. A plurality of requests from a plurality of client computers on the network stay in the input / output queue in the server computer. When each request is taken out from the input / output queue in the server computer, the requesting client computer is added to each request in advance. A message queuing method for sequentially extracting based on priority information,
Transmitting priority change information for changing the priority for a request from the client computer to the server computer via the network from the client computer;
Storing in the storage device the priority change information transmitted from the client computer in the server computer;
When fetching a request from the input / output queue in the server computer, a request to be fetched with priority is specified based on the priority information previously added to each request and the priority change information stored in the storage device. And the step of
A message queuing method, wherein the server computer retrieves requests from the input / output queue in an order based on the priority information and the priority change information. A message queuing system that sequentially retrieves a plurality of requests staying in an input / output queue based on priority information added in advance to each request by a requester,
Priority management means for registering priority change information notified from the request source of the request staying in the input / output queue,
Specifying means for specifying a request to be preferentially taken out based on the priority information previously added to each request and the priority change information registered by the priority management means when taking out requests from the input / output queue A message queuing system, wherein requests are taken out of an input / output queue in an order based on the priority information and the priority change information. A message queuing system that sequentially retrieves each request from a plurality of computers on a network staying in an input / output queue based on priority information added in advance to each request by a requesting computer,
Priority management means for registering priority change information notified via the network from the requesting computer of the request staying in the input / output queue,
Specifying means for specifying a request to be preferentially taken out based on the priority information added to each request in advance and the priority change information registered by the priority management means when taking out requests from the input / output queue A message queuing system, wherein requests are taken out of an input / output queue in an order based on the priority information and the priority change information.

Images