JP2013206041A - Communication system and load distribution processing apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a communication system and a load distribution processing apparatus for stably operating a function to be provided on a network.SOLUTION: The communication system includes a plurality of servers 31 to 34 and a load distribution processing apparatus 21, and the load distribution processing apparatus 21 includes a storage device 52 and a priority management device 51. The storage device 52 forms a queue as a data structure in which a storage area is formed, and the queue stores data with priority in the order of supply, and extracts the data with priority whose priority order is the highest from among the stored data with priority. When a fetch request is made, the priority management device 51 determines a server to which the processing of the data with priority is assigned from among the plurality of servers 31 to 34, supplies the data with priority to the determined server, and makes the determined server increase the usage of the number of thresholds set in the computer environment of the determined server.

Description

  The present invention relates to a communication system and a load distribution processing device, and in particular, assigns data to a server based on priority, and increases the usage of the number of threads provided in the computer environment of the assigned server, thereby increasing the usage on the network. The present invention relates to a communication system and a load distribution processing apparatus that stably operate the functions provided by the system.

Currently, a load distribution processing device is used as a device that assigns processing requests evenly in an information system to various processing requests on a network and efficiently executes the assigned processing (for example, Patent Document 1). reference).
When a processing request arrives, such a load distribution processing device distributes the requested processing to a plurality of servers on the network and executes transfer of the requested processing.
The distribution of the load, which is a requested process at this time, is executed based on a round robin method, a load balancing method according to the load status of the server, or the like.
Then, the server that has received the requested processing performs priority control according to the content of the requested processing, and sequentially executes the requested processing.

Japanese Patent No. 3964896 (paragraph [0003])

As described above, the conventional load distribution processing device does not execute the priority control, and the server that receives the requested process executes the priority control.
FIG. 5 is a diagram showing a schematic configuration of a conventional system. As shown in FIG. 5, the conventional load distribution processing device 201 transfers requested processing according to the load status of the servers 301 to 303.

For example, in the load distribution state 101, it is assumed that a process 351 having a high priority and a request having a high priority and a process 352 having a low priority and a request having a low priority are sequentially received. Since the server 301 is in a low load state, the next requested process 351 is transferred. Further, if the server 302 is in a high load state but the process 351 is transferred to the server 301 and the server 301 is evaluated as having a higher load, the next requested process 352 is transferred.
On the other hand, in the state 102 after load distribution, the server 301 is increasing in load, and the execution time of the processing 351 requested with high priority is longer than that in the state 101 during load distribution. Further, the load on the server 302 is decreasing, and the execution time of the processing 352 requested with low priority is less than that in the state 101 when the load is distributed.

In this way, as a result of the change in the load status of the servers 301 and 302 after transferring the requested processing, the completion of execution of the requested processing 351 with the high priority in the server 301 is completed in the server 302 which is another server. After the execution of the requested processing 352 of the low priority of, there may have been.
For this reason, there is a possibility that a state in which priority control cannot be accurately performed as a system may occur.

  Therefore, a communication system and a load distribution processing apparatus that can stably operate functions provided on a network have been desired.

  In the communication system according to the present invention, when priority-added data to which a priority indicating processing priority is added and a plurality of servers provided on the network are supplied from the outside, based on the priority, A load distribution processing device that determines a server to which processing of the data with priority is assigned from among a plurality of servers, and the load distribution processing device has the priority when the data with priority is supplied. A storage device that temporarily stores data, and a priority that manages the data with priority stored in the storage device based on the priority, and assigns processing of the data with priority to the plurality of servers The storage device forms a queue as a data structure that forms a storage area, and the queue includes the priority-added data in the order of supply. And the priority management device makes a request to retrieve the data with priority, from among the data with priority that has been stored, the data with priority having the highest priority. Until the storage device stores the prioritized data in order, and when the retrieval request is made, a server to which the processing of the prioritized data is assigned is determined from the plurality of servers, and the determined server The prioritized data fetched from the queue is supplied to the server, and the usage amount of the number of threads provided in the computer environment of the determined server is increased for the determined server.

  In the communication system of the present invention, the priority management device makes the retrieval request to the storage device, and the priority-added data retrieved from the queue of the storage device based on a use state of a computer environment of the server Is assigned to the server, and the prioritized data is supplied to the determined server.

  In the communication system of the present invention, the priority management device is set to one of a congested state and a free state as the use state, and the server in the use state is free from the plurality of servers. And processing of the data with priority taken out from the queue of the storage device.

  In the communication system according to the present invention, the storage device includes a plurality of the queues, and the allocation of the priority data to the server is performed in the queue as an execution ratio for executing the processing of the priority data. The priority management device is a division unit of processing that uses the computer environment of the server, and a plurality of threads provided in the computer environment are used with the use status of the plurality of threads, The processing of the data with priority is assigned based on the execution ratio.

  In the communication system of the present invention, the queue of the storage device is formed in a nested structure having a plurality of the queues therein.

  When load-priority data to which a priority indicating processing priority is added is supplied from the outside, the load distribution processing device of the present invention is configured to select the priority from a plurality of servers based on the priority. In the load distribution processing apparatus that selects a server that executes attached data and assigns a process to the selected server, the load distribution processing apparatus temporarily stores the prioritized data when the prioritized data is supplied. A storage device for storing, a priority management device that manages the data with priority stored in the storage device based on the priority, and assigns processing of the data with priority to the plurality of servers; And the storage device forms a queue as a data structure forming a storage area, and the queue stores and stores the prioritized data in the order of supply. Out of the data with priority, the priority data having the highest priority is extracted from the data with priority, and the priority management device stores the data in the storage device until making a request to extract the data with priority. When the prioritized data is stored in order and the retrieval request is made, a server to which the processing of the prioritized data is assigned is determined from the plurality of servers, and the prioritized data is stored in the determined server. At the same time, the amount of threads used in the computer environment of the determined server is increased for the determined server.

  In the load distribution processing device of the present invention, the priority management device makes the retrieval request to the storage device, and the priority retrieved from the queue of the storage device based on a use state of a computer environment of the server The server to which the attached data is assigned is determined, and the prioritized data is supplied to the determined server.

  In the load distribution processing device of the present invention, the priority management device is set to one of a congested state and a free state as the use state, and the use state is free among the plurality of servers. To the server, the processing of the data with priority taken out from the queue of the storage device is assigned.

  In the load distribution processing device according to the present invention, the storage device includes a plurality of the queues, and an allocation ratio in which the allocation of the priority data to the server is executed in the queue so that the processing of the priority data is executed. Is set in advance, and the priority management device is a unit of processing for using the computer environment of the server, and the usage status of the plurality of threads is included in the plurality of threads provided in the computer environment. The processing of the data with priority is assigned based on the execution ratio.

  In the load distribution processing device of the present invention, the queue of the storage device is formed in a nested structure having a plurality of the queues therein.

  The present invention stably operates functions provided on a network by allocating data to servers based on priorities and increasing the amount of threads used in the computer environment of the allocated servers. Has the effect of being able to.

It is a figure which shows schematic structure of the communication system 11 in Embodiment 1 of this invention. It is a sequence which shows operation | movement of the communication system 11 in Embodiment 1 of this invention. It is a figure which shows schematic structure of the communication system 11 in Embodiment 2 of this invention. It is a sequence which shows operation | movement of the communication system 11 in Embodiment 2 of this invention. It is a figure which shows schematic structure of the system of a prior art.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

Embodiment 1 FIG.
<Description of configuration>
FIG. 1 is a diagram showing a schematic configuration of a communication system 11 according to Embodiment 1 of the present invention. As shown in FIG. 1, the communication system 11 is connected to a large number of clients 3 via a network 1 and responds to processing requests from a large number of clients 3.

  The communication system 11 transfers processing requested from a large number of clients 3 to servers 31 to 34, which will be described later, and distributes the load in the entire system, and processing transferred from the load distribution processing device 21. Servers 31 to 34 to be executed, and a computer environment 41 for executing the processes transferred to the servers 31 to 34 are provided.

The computer environment 41 provides physical computer resources to the servers 31 to 34.
The computer environment 41 includes a thread 71.
The thread 71 means a set of threads composed of a plurality of threads.
Each thread in the thread 71 is a division unit of processing in which the servers 31 to 34 use the computer environment 41.
In addition to the thread 71, the computer environment 41 includes, for example, a CPU (Central Processing Unit).

Next, the relationship between the servers 31 to 34 and the computer environment 41 will be described.
Physically, the servers 31 to 34 and the computer environment 41 are mounted on, for example, a rack mount server.
The computer environment 41 is a computer resource that the servers 31 to 34 can use in common. For example, it is assumed that the servers 31 to 34 have a predetermined number of threads to be used for each server.
For example, if a total of 10 threads can be provided to the servers 31 to 34 as the computer environment of the entire system, the server 31 has 4 threads, the server 32 has 2 threads, the server 33 has 2 threads, and the server 34 has threads. It is assumed that two can be assigned.
For example, under the above assumption, when the server 31 currently uses two threads and supplies data with priority to the server 31, two more threads can be allocated to the server 31.
The computer environment 41 may be composed of a plurality of servers. In this case, the computer resources used by one or a plurality of servers among the servers 31 to 34 may be different. However, since the load distribution processing device 21 performs load control in units of threads, it is the same as when the computer resources are the same. Can behave. By adopting such a configuration, an effect of improving scalability and fault tolerance can be obtained.

The servers 31 to 34 have a functional configuration that provides some service, that is, processing in response to a request from the client 3, that is, the request.
For example, in the case of a file server, a stored file, that is, data is provided.
In the case of a print server, it provides a printing process to a printer.
In the case of a web server, data such as an HTML (Hyper Text Markup Language) file or an image file that constitutes a web page is provided.
In addition, one physical computer may include a plurality of servers such as a printer server and a web server.
Further, for example, one physical computer may include one file server, one computer may have the function of one server, and a plurality of computers may be provided for each server.

In addition, although the server 31-34 showed about the example which provided four servers here, it is not limited to this. For example, two servers or 100 servers may be used.
Moreover, although the servers 31-34 here are web servers, for example, they are not limited to this. For example, it may be a business support server that performs payment processing, a mail server that processes mail, or the like. It may also be a SIP server that performs call control.

Next, details of the servers 31 to 34 will be described.
The servers 31 to 34 do not execute priority control for processing requests from the client 3.
The servers 31 to 34 have queues 35 to 38 for managing the usage state of the computer environment 41, that is, the usage state of the thread 71 provided in the computer environment 41.

The queues 35 to 38 are one of data structures that form a storage area.
The queues 35 to 38 hold data in a first-in first-out (FIFO) list structure.
That is, the queues 35 to 38 are formed by queues.
In the queues 35 to 38, when data is taken out from the queue, the data is taken out in order from the data put in first.
In the queues 35 to 38, a threshold value Nb for identifying a state where the usage state of the thread 71 is congested, that is, a state where the usage state of the thread 71 is Busy is set.
In the queues 35 to 38, a threshold value Na is set for identifying a state where the usage state of the thread 71 is vacant, that is, a state where the usage state of the thread 71 is Active.
Details of the relationship between the servers 31 to 34 and the usage state of the thread 71 will be described later.

Here, when any one of the queues 35 to 38 is indicated, it is referred to as a queue 39.
The size of the entire queue 39 is different for each server.
For example, the total number of threads 71 that can be provided by the computer environment to the servers 31 to 34 is 100. At this time, it is assumed that 30 threads are allocated to the server 31, 20 threads are allocated to the server 32, 10 threads are allocated to the server 33, and 40 threads are allocated to the server 34 as usable threads.
In the above assumption, the available thread size relationship is server 34> server 31> server 32> server 33. Therefore, the size relationship of the queue 39 is queue 38> queue 35> queue 36> queue 37.

Next, specific examples of the threshold value Nb and the threshold value Na will be described.
For example, it is assumed that the processing contents requested from the client 3 are sequentially stored in the queue 39 and reach 50% of the entire queue 39. That is, it is assumed that the processing content in standby has reached 50% of the queue 39.
At this time, if the threshold Nb is set to 50% of the queue 39, it is assumed that the usage state of the thread 71 is Busy. When the usage state of the thread 71 is Busy, the server 39 does not accept transfer of the processing request of the client 3 by the load distribution processing device 21.
Further, after this, it is assumed that the processing content in standby has reached 30% of the queue 39.
At this time, if the threshold value Na is set to 30% of the queue 39, the use state of the thread 71 is assumed to be Active. When the usage state of the thread 71 is Active, the server 39 accepts the transfer of the processing request of the client 3 by the load distribution processing device 21.
As described above, the queue 39, the threshold value Nb, and the threshold value Na allow the number of usable threads to be increased, the servers 31 to 34 are in a high load state, and a time lag occurs until the next process is accepted. Try to prevent.

Next, functions of the load distribution processing device 21 will be described.
When the priority distribution data to which the priority indicating the priority of processing is added is supplied from the outside such as the client 3, the load distribution processing device 21 selects the server 31 to 34 based on the priority. A server to which processing of data with priority is assigned is determined.

Next, the configuration of the load distribution processing device 21 will be described.
The load distribution processing device 21 includes a priority management device 51 that manages the priority of processing requests of the client 3 and a storage device 52 that temporarily stores processing requests of the client 3.
Here, it is assumed that the processing request of the client 3 includes data with priority to which a priority indicating the processing priority is added in advance.
As data with such priority, for example, when the servers 31 to 34 are SIP servers, emergency telephone number data such as 119 or 110 is provided in Japan, and 911 emergency telephone data is provided in the United States. If it is England, it is 999 emergency call data.
In addition, for example, data with priority is data such as a request for approval when the servers 31 to 34 are business support servers such as companies.
For example, when the servers 31 to 34 are mail servers, the data with priority is a flag indicating the degree of urgency added to the mail.

For identification of such data that requires urgency, for example, application data included in the processing request data may be decrypted.
In addition, a flag for identifying data that requires urgentness may be newly provided in the most significant bit of the user data, and an application header may be provided in the user data as application data, and the application data may be decoded.
In short, the level of priority is based on the request content, the request source, the label attached to the request, and the like.
In any case, the mounting form for adding the priority is not particularly limited.

  When the data with priority is supplied, the storage device 52 temporarily stores the data with priority.

Next, details of the storage device 52 will be described.
The storage device 52 forms a queue 61 as a data structure that forms a storage area.
The queue 61 holds data in a first-in first-out (FIFO) list structure.
That is, the queue 61 is formed by a queue.
When data is taken out from the queue 61, the queue 61 is taken out in order from the data put in first.
The queue 61 stores data with priority in the order of supply, and retrieves the data with priority having the highest priority among the stored data with priority.

  When the data with priority is supplied, the priority management device 51 manages the data with priority stored in the storage device 52 that temporarily stores the data with priority based on the priority. To 34 are assigned data processing with priority.

Next, details of the priority management apparatus 51 will be described.
The priority management device 51 causes the storage device 52 to store the data with priority in order until the data with priority is taken out.
When the priority management apparatus 51 requests the storage device 52 to retrieve data with priority, the priority management apparatus 51 determines a server to which processing of data with priority is assigned from among the servers 31 to 34, and assigns priority to the determined server. Supply the attached data.
When the priority management apparatus 51 supplies data with priority to the determined server, the priority management apparatus 51 increases the usage amount of the thread 71 provided in the computer environment of the server.

  Specifically, the priority management device 51 selects the data with priority retrieved from the queue 61 based on the usage state of the computer environment of each server among the servers 31 to 34, for example, the number of threads 71 used. A server to be allocated is determined, and data with priority is supplied to the determined server.

More specifically, the priority management apparatus 51 includes servers 31 to 34 when there is a server that uses a thread 71 that is in a busy state among the servers 31 to 34, for example, the usage state of the computer environment of each server. 34, the processing of the data with priority taken out from the queue 61 is assigned to a server having a thread 71 available.
That is, the priority management apparatus 51 assigns the processing of the data with priority retrieved from the queue 61 to a server among the servers 31 to 34 where the number of usable threads 71 is still free.
In other words, when the usage state of the thread 71 indicates Busy by the threshold value Nb, the priority management apparatus 51 does not assign processing of data with priority to a server in that state.
On the other hand, when the usage state of the thread 71 indicates Active due to the threshold value Na, the priority management apparatus 51 assigns processing of data with priority to the server in that state.

Next, the relationship between the servers 31 to 34 and the usage state of the thread 71 will be described.
As described above, thread allocation is determined for each server. In addition, the server thread 31 uses the threshold value Nb and threshold value Na of the queue 35 for the server 31, the threshold value Nb and threshold value Na for the queue 36 for the server 32, the threshold value Nb and threshold value Na for the queue 37 for the server 33, and The threshold value Nb and the threshold value Na of the queue 38 are replaced with the values to be judged.
Therefore, Busy or Active in the usage state of the thread 71 here does not mean that the total number of threads that the computer environment 41 can provide to the servers 31 to 34 is Busy or Active.
The busy state of the thread 71 used here means, for example, when the threshold value Nb of the queue 35 reaches 40% when the threshold value Nb of the queue 35 is set to 40%. Since the queue 35 is managed by the server 31, at this time, the priority management device 51 does not assign processing of data with priority to the server 31.
Further, the active state of the thread 71 used here means, for example, when the threshold value Na of the queue 36 is set to 15% when the threshold value Na of the queue 36 is set to 15%. Since the queue 36 is managed by the server 32, at this time, the priority management apparatus 51 assigns processing of data with priority to the server 32.

<Description of operation>
FIG. 2 is a sequence showing the operation of the communication system 11 according to the first embodiment of the present invention.
When the execution request from the client 3 reaches the load distribution processing device 21, the execution request is stored in the queue 61.
The execution request stored in the queue 61 is taken out by the priority management device 51 and transferred to an executable server.
Specifically, the priority management device 51 makes a request for taking out priority-added data as an execution request to the storage device 52 in which the queue 61 is formed, and obtains data with priority as a take-out request reply. .

Next, the priority management apparatus 51 requests the servers 31 to 34 for the usage status of the computer environment 41 and acquires the threshold value Nb and the threshold value Na as a response to the usage status request of the computer environment 41.
At this time, for example, the threshold Na of the queue 36 of the server 32 is Active, the threshold Nb of the queue 35 of the server 31, the threshold Nb of the queue 37 of the server 33, and the threshold Nb of the queue 38 of the server 34 are Busy. To do.
The usage status of the servers 31 to 34 is confirmed in advance by holding the status on the load distribution processing device 21 by an inquiry from the load distribution processing device 21 or self-reporting from the servers 31 to 34 and confirming it. It may be a method.

In this case, the usage state of the computer environment 41 of the server 32 is empty, and the usage state of the computer environment 41 of the server 31, the server 33, and the server 34 is congested.
Therefore, the priority management apparatus 51 instructs the server 32 to assign processing of data with priority.

The priority management device 51 increases the number of threads within the usable range of the number of threads assigned to the server 32 when assigning processing of data with priority to the server 32. For example, when the number of usable threads assigned to the server 32 is 130 and the number of currently used threads is 30, the priority management apparatus 51 includes 50 threads out of the remaining 100 threads. Let the process be assigned.
In this way, the priority management apparatus 51 increases the amount of threads used.

  Further, the priority management device 51 requests the server 32 to perform further processing even if the computer 32 of the computer environment 41 is still in use after the server 32 has been assigned the processing of data with priority. May be kept to the minimum necessary. For example, if the number of transfers per server is reduced to 1, the response quality can be equalized even if the operation cannot be executed in a multiplexed manner.

  Also, the priority management device 51, when one of the servers 31 to 34 in which the usage state of the computer environment 41 is active occurs at an early stage, the data with priority from the queue 61 with the highest priority. Are taken in order, become Active, and are transferred to the server.

<Description of effects>
As described above, by performing priority control before performing load distribution, it is possible to prevent occurrence of inversion of processing priority as a system that has occurred in the prior art.
In other words, the function provided on the network can be stably operated by assigning data to the server based on the priority and increasing the usage amount of the number of threads provided in the computer environment of the assigned server. .
In other words, the quality of QoS (Quality of Service) as a system can be improved.

<Description of usage pattern>
If the above configuration is applied to an HTTP request, it becomes possible to preferentially process VIP user priority processing and DB (database) update processing as a whole system.
Moreover, if the above configuration is applied to a SIP request, an emergency contact can be preferentially connected.

  As described above, in the first embodiment, when the data with priority to which the plurality of servers 31 to 34 provided on the network and the priority indicating the processing priority are added is supplied from the outside, A load distribution processing device 21 that determines a server to which processing of data with priority is assigned from a plurality of servers 31 to 34 based on the priority, and the load distribution processing device 21 is supplied with data with priority. Storage device 52 that temporarily stores data with priority, and data with priority stored in storage device 52 based on the priority, and prioritizes multiple servers 31-34. A priority management device 51 for allocating data processing, and the storage device 52 forms a queue 61 as a data structure forming a storage area, and the queue 61 is supplied The priority data is stored in the data, and the priority-priority data having the highest priority is extracted from the stored data with priority, and the priority management device 51 makes a request to retrieve the data with priority. Until the data with priority is stored in the storage device 52 in order and a retrieval request is made, a server to which processing of data with priority is assigned is determined from the plurality of servers 31 to 34, and the determined server is determined. By supplying the data with priority retrieved from the queue 61 and increasing the usage of the number of threads provided in the computer environment of the determined server for the determined server, the function provided on the network is stabilized. Can be operated automatically.

Embodiment 2. FIG.
The difference from the first embodiment is that a plurality of queues 61 and 62 are mounted in the storage device 52 and there is an environment setting file 63 in which the setting contents of the queues 61 and 62 are stored.
In the second embodiment, items that are not particularly described are the same as those in the first embodiment, and the same functions and configurations are described using the same reference numerals.
Further, the description of the same functions, configurations, and operations as those in Embodiment 1 is omitted.

<Description of configuration>
FIG. 3 is a diagram showing a schematic configuration of the communication system 11 according to Embodiment 2 of the present invention.
As illustrated in FIG. 3, the priority management device 51 includes a queue 61 and a queue 62 in the storage device 52.
Here, an example in which the two queues 61 and 62 are mounted will be described, but the number of queues mounted is not limited to this.
Further, the queue 61 may have a nested structure with a plurality of queues.
Further, the queue 62 may have a nested structure in which a plurality of queues are further provided.

The priority management device 51 includes an environment setting file 63 in the storage device 52.
The environment setting file 63 is so-called config data, and stores various information set by an administrator of the communication system 11, a user of the communication system 11, and the like.

For example, in the queue 61 and the queue 62, distributions for assigning processes to the servers are set at different rates, and such setting contents are stored in the environment setting file 63.
Specifically, in the queue 61 and the queue 62, a distribution for allocating data with priority to the servers 31 to 34 is set in advance as an execution ratio for executing processing of data with priority, and the setting content is an environment setting file. 63.
For example, when the queue 61: queue 62 is set as 3: 1, the servers 31 to 34 have an execution ratio of the processing stored in the queue 61 and the processing stored in the queue 62 of 3: 1. Become. That is, in this case, the contents of the queue 61 are preferentially assigned.
Therefore, by determining the execution ratio, it becomes possible to control the degree of processing allocation to the servers 31 to 34.

That is, the priority management apparatus 51 assigns the processing of the data with priority to the servers 31 to 34 based on the thread usage status and the execution ratio.
Specifically, it is assumed that the processing content stored in the queue 61 is settlement processing that requires urgency, and the processing content stored in the queue 62 is month-end processing that does not require urgency.
In this case, by setting the execution ratio of the queue 61 and the queue 62 to 9: 1, for example, both processes are stably executed.
More specifically, the priority management apparatus 51 assigns the processing stored in the queue 61 to the servers 31 to 34 with priority over the processing stored in the queue 62, and further, the priority in the queue 61 itself. Are assigned to the servers 31 to 34 with priority.
At the same time, the priority management apparatus 51 assigns processes to the servers 31 to 34 at a certain rate even for the processes stored in the queue 62.
As a result, the servers 31 to 34 can process both the settlement process and the month end process without delay.
That is, the priority management apparatus 51 can control the processing amount according to the attribute of the request by managing a plurality of queues.

<Description of operation>
FIG. 4 is a sequence showing an operation of the communication system 11 according to the second embodiment of the present invention.
The priority management device 51 requests the storage device 52 to retrieve data with priority, and also requests information on the environment setting file.
As in the communication system 11 in the first embodiment described with reference to FIG. 2, the usage status of the servers 31 to 34 is confirmed in advance by an inquiry from the load distribution processing device 21 or a self-report from the servers 31 to 34. A method of holding the status on the load distribution processing apparatus 21 and confirming it may be used.

Next, the priority management apparatus 51 acquires the execution ratio of the queue 61 and the queue 62 from the environment setting file. At this time, the execution ratio between the queue 61 and the queue 62 is assumed to be 2: 1.
Thereafter, the priority management apparatus 51 determines that the server 32 is active, and transmits the data with priority in the queue 61 to the server 32.
Further, thereafter, the priority management apparatus 51 assigns processing to the server that has become active early while maintaining the execution ratio of the queue 61 and the queue 62.

<Description of effects>
As described above, by performing priority control before performing load distribution, it is possible to prevent occurrence of inversion of processing priority as a system that has occurred in the prior art.
That is, functions provided on the network can be stably operated.
Also, by providing a plurality of queues in the load distribution processing device 21, it becomes possible to control the processing amount according to the attribute of the request.
In other words, the quality of QoS (Quality of Service) as a system can be improved.

<Description of usage pattern>
With the above configuration, it is possible to mix a plurality of tasks, assign all the task servers during a busy period of a particular task, and execute other tasks without being affected as necessary.

  As described above, in the second embodiment, the storage device 52 has a plurality of queues, and the allocation of the priority data to the servers 31 to 34 in the queue causes the priority data processing to be executed. The execution ratio is set in advance, and the priority management apparatus 51 is a division unit of processing that uses the computer environment 41 of the servers 31 to 34, and a plurality of threads provided in the computer environment 41 are assigned to the plurality of threads. By assigning priority data processing based on thread usage and execution ratio, functions provided on the network can be stably operated.

  1 network, 3 client, 11 communication system, 21 load distribution processing device, 31, 32, 33, 34 server, 41 computer environment, 51 priority management device, 52 storage device, 35, 36, 37, 38, 39, 61 , 62 queue, 63 environment setting file, 71 thread, 101 load distribution status, 102 load distribution status, 201 conventional load distribution processing device, 301, 302, 303 server, 311, 312, 313 queue, 351 high Request processing with priority, 352 Request processing with low priority.

Claims (10)

A plurality of servers provided on the network;
A server that assigns processing of the data with priority among the plurality of servers based on the priority when the data with priority to which the priority indicating the priority of processing is added is supplied from the outside. And a load balancing processing device for determining
The load distribution processing device includes:
A storage device for temporarily storing the data with priority when the data with priority is supplied;
A priority management device that manages the data with priority stored in the storage device based on the priority, and assigns processing of the data with priority to the plurality of servers;
Have
The storage device
A queue is formed as a data structure that forms a storage area.
The queue is
Storing the prioritized data in the order in which they are supplied, and out of the stored prioritized data, the priority-priority data having the highest priority is extracted from the prioritized data;
The priority management device includes:
Until the data with priority is requested to be taken out, the storage device stores the data with priority in order,
When the retrieval request is made, a server to which the priority data processing is assigned is determined from the plurality of servers, and the priority data retrieved from the queue is supplied to the determined server, and the determination is performed. A communication system characterized by increasing the usage amount of threads provided in the determined computer environment of the server. The priority management device includes:
Making the retrieval request to the storage device;
A server to which the data with priority extracted from the queue of the storage device is assigned is determined based on a use state of a computer environment of the server, and the data with priority is supplied to the determined server. The communication system according to claim 1. The priority management device includes:
As the use state, either one of the congested state and the vacant state is set,
3. The communication system according to claim 2, wherein, among the plurality of servers, processing of the data with priority taken out from the queue of the storage device is assigned to a server in which the usage state is vacant. The storage device includes a plurality of the queues,
In the queue, a distribution for allocating the data with priority to the server is set in advance as an execution ratio for executing processing of the data with priority.
The priority management device includes:
A division unit of processing that uses the computer environment of the server, and a plurality of threads provided in the computer environment, the usage status of the plurality of threads and the execution ratio of the data with priority 4. The communication system according to claim 1, wherein a process is assigned. The communication system according to claim 4, wherein the queue of the storage device is formed in a nested structure having a plurality of the queues therein. When priority-added data to which a priority indicating processing priority is added is supplied from the outside, a server that executes the data with priority is selected from a plurality of servers based on the priority. In the load distribution processing device that assigns processing to the selected server,
The load distribution processing device includes:
A storage device for temporarily storing the data with priority when the data with priority is supplied;
A priority management device that manages the data with priority stored in the storage device based on the priority, and assigns processing of the data with priority to the plurality of servers;
Have
The storage device
A queue is formed as a data structure that forms a storage area.
The queue is
Storing the prioritized data in the order in which they are supplied, and out of the stored prioritized data, the priority-priority data having the highest priority is extracted from the prioritized data;
The priority management device includes:
Until the data with priority is requested to be taken out, the storage device stores the data with priority in order,
When the retrieval request is made, a server to which the priority data processing is assigned is determined from the plurality of servers, and the priority data retrieved from the queue is supplied to the determined server, and the determination is performed. A load distribution processing device, wherein the usage amount of the number of threads provided in the computer environment of the determined server is increased for the determined server. The priority management device includes:
Making the retrieval request to the storage device;
A server to which the data with priority taken out from the queue of the storage device is assigned is determined based on a use situation of a computer environment of the server, and the data with priority is supplied to the determined server. The load distribution processing device according to claim 6. The priority management device includes:
As the use state, either one of the congested state and the vacant state is set,
8. The load balancing process according to claim 7, wherein, among the plurality of servers, the process with the priority data fetched from the queue of the storage device is assigned to a server in which the usage state is vacant. apparatus. The storage device includes a plurality of the queues,
In the queue, a distribution for allocating the data with priority to the server is set in advance as an execution ratio for executing processing of the data with priority.
The priority management device includes:
A processing unit that uses the computer environment of the server, and a plurality of threads provided in the computer environment are processed based on the usage status of the plurality of threads and the execution ratio. The load distribution processing device according to claim 6, wherein the load distribution processing device is assigned. The load distribution processing device according to claim 9, wherein the queue of the storage device is formed in a nested structure having a plurality of the queues therein.

Images