JP6204287B2 - Distributed processing method, processing server, and program - Google Patents

Description

  The present invention relates to a technique for processing a request from a user terminal by sharing it with a plurality of processing servers.

  As a technique for performing distributed processing for determining a processing server in charge of a request from a user terminal, for example, there is a distributed data management method in which processing is distributed using a consistent hash method (Patent Document 1). In this distributed data management method, a consistent hash method is used, and a processing server is allocated on the virtual hash space. When the distributed data management device receives a request (service request signal) from a user terminal, the request is identified. The value in the virtual hash space is calculated based on the key to be determined, and the processing server (service processing server) responsible for the response (service processing) to the request is determined from the calculated value in the virtual hash space. Distributed processing is realized by transferring the request to the processing server.

  The distributed processing (D (Dispatch) signal processing) and service processing (PS (Processing) signal processing) can be executed in a superimposed manner on one machine (processing server). By superimposing and executing on one machine, if the service processing server is its own processing server as a result of D signal processing, there is no need for inter-machine communication, and the cost for that communication is reduced. There is an advantage that can be.

JP 2011-95976 A

  In a situation where distributed processing and service processing are superimposed on one processing server, it should be avoided that resources for distributed processing (CPU (Central Processing Unit), memory, etc.) compress service processing resources. . This is because even if distribution can be performed by D signal processing, if the important service processing is delayed, the number of requests retained on the processing server side will increase, leading to an increase in resource usage, or the response of the processing server This is because the time may be increased and the influence on the user terminal side may occur.

  In general, in a communication server, software is designed based on a so-called outgoing side priority concept that prioritizes transmission processing over signal reception processing. Therefore, it is desirable to design software with priority on the outgoing side for processing servers that perform distributed processing.

  Therefore, an object of the present invention is to avoid a delay in service processing due to an increase in distributed processing in a processing server that performs distributed processing and service processing in parallel.

  In order to solve the above-mentioned problem, the present invention determines a distributed processing for determining a processing server in charge of the request when each of the plurality of processing servers receives a request from a user terminal, and determines the processing server in charge. A distributed processing method by the processing server in a processing system that performs service processing responding to the request in parallel when the processing server performs the distributed processing, the distributed processing unit performing the distributed processing, and the service processing A service processing unit that performs resource processing, resource usage information indicating resource usage status related to the distributed processing, a storage unit that stores a predetermined threshold value related to the resource usage status information, and a resource monitoring function unit The resource monitoring function unit refers to the resource usage status information and the predetermined threshold value, and If the source usage information is greater than the predetermined threshold value, notification to the dispersion pause, maintainer terminal processing, and characterized by performing additional resources, the at least one.

  According to this, in a processing server that performs distributed processing and service processing in parallel, when the resource usage status information exceeds a predetermined threshold, suspension of distributed processing, notification to a maintenance person terminal, and resource By executing at least one of the above, it is possible to avoid a delay in service processing due to an increase in distributed processing.

  In addition, the present invention is characterized in that the resource usage status information is information on a CPU (Central Processing Unit) or memory usage rate related to distributed processing.

  According to this, by using information on the CPU or memory usage rate related to the distributed processing, which increases directly as a resource processing status information, the service processing delay due to the increase in the distributed processing can be reduced. This can be avoided more reliably.

  In addition, the present invention is characterized in that the resource usage status information is information on the number of distributed processes per unit time.

  According to this, by using the information on the number of distributed processes per unit time, which directly leads to a delay in service processing as the resource usage status information increases, the delay in service processing due to the increase in distributed processing is further increased. It can be avoided reliably.

  In addition, the present invention can be realized as a program for causing a computer to function as the processing server. Thereby, a computer can be functioned as the processing server.

  According to the present invention, in a processing server that performs distributed processing and service processing in parallel, a delay in service processing due to an increase in distributed processing can be avoided.

It is a whole block diagram containing the processing server of this embodiment. (A) is a figure showing an example of resource use status information. (B) is a figure showing an example of threshold information. It is the sequence diagram which showed the flow of processes, such as a process server in this embodiment. (A) is a figure showing other examples of resource use status information. (B) is a figure showing other examples of threshold information.

  EMBODIMENT OF THE INVENTION The form (embodiment) for implementing this invention is demonstrated referring drawings. As shown in FIG. 1, the processing system 100 includes a processing server 1 and a plurality of other processing servers 3. Note that the processing server 1 and a plurality of other processing servers 3 are equivalent, but here, the processing server 1 is considered. Further, when the processing server 1 and a plurality of other processing servers 3 are expressed together, they are described as “processing servers 1 and 3”. In addition, when referring to one of “processing servers 1 and 3”, “processing server” may be described without any reference.

  In the processing system 100, when one of the processing servers 1 and 3 receives a request from the user terminal 2, the received processing server performs distributed processing for determining the processing server in charge of the request among the processing servers 1 and 3. And the determined processing server performs service processing in response to the request. That is, each of the processing servers 1 and 3 performs distributed processing and service processing in parallel.

  The processing server 1 includes a D signal reception function unit 11, a D signal processing function unit 12, a PS signal reception function unit 13, a PS signal processing function unit 14, and a transmission function unit 15 that are realized by a CPU (Central Processing Unit), a memory, and the like. , A resource monitoring function unit 16 and a storage unit 17 realized by an HDD (Hard Disk Drive), a flash memory, or the like.

  The storage unit 17 stores resource usage status information 18 and threshold information 19 (predetermined threshold). The storage unit 17 stores a program for causing a computer to function as the processing server 1. With this program, the computer can function as the processing server 1.

The D signal reception function unit 11 extracts a request signal (D signal) from the user terminal 2 from the reception queue.
The D signal processing function unit 12 (distributed processing unit) analyzes the D signal and performs distributed processing for determining a processing server in charge, for example, by a consistent hash method.

The PS signal reception function unit 13 takes out a signal (PS signal) from another processing server 3 from the reception queue.
The PS signal processing function unit 14 (service processing unit) analyzes the PS signal and performs service processing.

The transmission function unit 15 transmits a signal to the processing server determined by the D signal processing function unit 12 and the user terminal 2.
The resource monitoring function unit 16 refers to the resource usage status information 18 and the threshold information 19, and when the resource usage status information 18 exceeds the value indicated by the threshold information 19, the distributed processing is temporarily stopped, and the maintenance person terminal 4 And / or resource addition (details will be described later).

  As shown in FIG. 2A, the storage unit 17 stores information on usage, CPU usage (%), and memory usage (%) for each thread ID (IDentifier) as resource usage status information 18. Stored. The resource usage status information 18 is sequentially updated by the D signal processing function unit 12 and the PS signal processing function unit 14.

  As shown in FIG. 2B, the storage unit 17 includes, as threshold information 19, for example, an overall CPU usage rate threshold, a CPU usage rate threshold for a D signal processing thread, an overall memory usage rate threshold, and a D signal processing thread. Each information of the memory usage rate threshold value for use is preset.

  Next, a processing flow of the processing server 1 and the like in the present embodiment will be described. As shown in FIG. 3 (see also FIG. 1), the user terminal 2 transmits a request signal (D signal, request) (S1), and the processing server 1 receives the request signal.

In the processing server 1, the D signal reception function unit 11 receives a request signal from the user terminal 2, and transfers the request signal to the D signal processing function unit 12 (S2).
When receiving the request signal, the D signal processing function unit 12 generates a thread (S3), and notifies the resource monitoring function unit 16 of the thread ID (S4). In addition, the D signal processing function unit 12 updates the resource usage status information 18 in the storage unit 17.

Next, the D signal processing function unit 12 performs distributed processing for determining a processing server in charge of the request signal (S5), and sends the request signal to the determined processing server via the transmission function unit 15. (PS signal) is transmitted (S6, S7).
The processing server (other processing server 3) that has received the request signal (PS signal) performs service processing in response to the request signal.

  The PS signal reception function unit 13 of the processing server 1 receives another request signal (D signal) from the user terminal 2 and receives a request signal (PS signal) from the other processing server 3 that has finished the distributed processing. (S8), the request signal is transferred to the PS signal processing function unit 14 (S9).

  When receiving the request signal, the PS signal processing function unit 14 generates a thread (S10), and notifies the resource monitoring function unit 16 of the thread ID (S11). Further, the PS signal processing function unit 14 updates the resource usage status information 18 in the storage unit 17.

  Next, the PS signal processing function unit 14 performs service processing related to the request signal (PS signal) (S12), and sends a response signal to the user terminal 2 that has issued the request signal via the transmission function unit 15. Is transmitted (S13, S14).

Further, the resource monitoring function unit 16 starts periodically and collects the resource usage status information 18 from the storage unit 17 using the thread ID received in S4 and S11 (S15).
Next, the resource monitoring function unit 16 determines the total CPU usage rate, the CPU usage rate for the D signal processing thread, the total memory usage rate, and the memory usage rate for the D signal processing thread based on the collected resource usage status information 18. And the respective values are the total CPU usage threshold, the CPU usage threshold for the D signal processing thread, the total memory usage threshold, and the memory usage threshold for the D signal processing thread stored in the threshold information 19. It is determined whether it exceeds (S16). If Yes, the process proceeds to S17, and if No, S17 is skipped.

  In S <b> 17, the resource monitoring function unit 16 stops the distributed processing, notifies the maintenance person terminal 4, and adds a resource (for example, a processing server) so that the resource for distributed processing does not press on the resource for service processing. Add at least one of the following.

If the distributed processing is temporarily stopped, it is possible to directly avoid the situation where the resources for the distributed processing press the resources for the service processing.
Further, if the maintenance person terminal 4 is notified, the maintenance person can take effective measures.
Further, if the resource is added, the usage rate of the resource for distributed processing is naturally reduced, and therefore, the situation where the resource for distributed processing presses the resource for service processing can be avoided.

  Note that when the distributed processing is paused, the distributed processing is resumed when a condition such as a predetermined time has elapsed or a value that has exceeded the threshold value has fallen below the threshold value is satisfied. That's fine.

  Receiving the notification, the maintenance person terminal 4 displays information regarding the value exceeding the threshold value on the display unit 31. The maintenance person can take measures by viewing this display.

  In this way, according to the processing server 1 of the present embodiment, when the distributed processing and the service processing are performed in parallel, the resource usage status information 18 (D signal processing, in particular, regarding the distributed processing among the resource usage status information 18). When the CPU usage rate for threads and the memory usage rate for D signal processing threads) exceed predetermined thresholds, at least the suspension of distributed processing, notification to the maintenance person terminal 4, and addition of resources are at least By executing either one, it is possible to avoid a delay in service processing due to an increase in distributed processing.

  Further, as the resource usage status information 18, information on the CPU or memory usage rate related to the distributed processing that directly leads to a delay in the service processing if used increases, so that the delay in the service processing due to the increase in the distributed processing is more reliably ensured. Can be avoided.

  Next, another example of resource usage status information and threshold information will be described. As shown in FIG. 4A, the number of processes per unit time of a signal (D signal) accompanied by distributed processing can be used as the resource usage status information 18a. In this case, as shown in FIG. 4B, a threshold value for the number of processes per unit time of a signal (D signal) accompanied by distributed processing is set in advance as the threshold information 19a.

  When the resource monitoring function unit 16 exceeds the value indicated by the threshold information 19a in the resource usage status information 18a, at least the suspension of the distributed processing, the notification to the maintenance person terminal 4, and the addition of the resource are at least Either one should be executed.

According to this, by using information on the number of processes per unit time of a signal accompanied by distributed processing, which directly leads to a delay in service processing as the resource usage status information increases, service processing due to increased distributed processing is performed. This delay can be avoided more reliably.
Note that when the distributed processing is temporarily stopped, for example, the distributed processing may be resumed when the unit time has elapsed thereafter.

Although description of this embodiment is finished above, the aspect of the present invention is not limited to these.
For example, the numbers such as “50.0%” shown in FIG. 2B are examples, and other numbers may be adopted.

Also, the timing of performing the processing of S15 to S17 in FIG. 3 is not limited to “periodic”, but is another timing, for example, every time the resource monitoring function unit 16 receives a thread ID notification a predetermined number of times. Also good.
In addition, about a concrete structure, it can change suitably in the range which does not deviate from the main point of this invention.

DESCRIPTION OF SYMBOLS 1 Processing server 2 User terminal 3 Other processing server 4 Maintenance person terminal 11 D signal receiving function part 12 D signal processing function part 13 PS signal receiving function part 14 PS signal processing function part 15 Transmission function part 16 Resource monitoring function part 17 Storage Unit 18, 18a Resource usage status information 19, 19a Threshold information 31 Display unit 100 Processing system

Claims (7)

A plurality of processing servers, each of which receives a request from a user terminal, a distributed process for determining a processing server in charge of the request, and a service process for responding to the request when determined as the processing server in charge , A distributed processing method by the processing server in a processing system for performing in parallel,
The processing server
A distributed processing unit for performing the distributed processing;
A service processing unit for performing the service processing;
A storage unit that stores resource usage information indicating resource usage related to the distributed processing, and a predetermined threshold value related to the resource usage information;
A resource monitoring function unit,
The resource monitoring function unit
Referring to the resource usage status information and the predetermined threshold value, when the resource usage status information exceeds the predetermined threshold value, the distributed processing is temporarily stopped, the notification to the maintenance person terminal, and the addition of the resource A distributed processing method characterized by executing at least one of the following. The distributed processing method according to claim 1, wherein the resource usage status information is information on a CPU (Central Processing Unit) or a memory usage rate related to the distributed processing. The distributed processing method according to claim 1, wherein the resource usage status information is information on the number of the distributed processes per unit time. A plurality of processing servers, each of which receives a request from a user terminal, a distributed process for determining a processing server in charge of the request, and a service process for responding to the request when determined as the processing server in charge , The processing server in a processing system for performing in parallel,
A distributed processing unit for performing the distributed processing;
A service processing unit for performing the service processing;
A storage unit that stores resource usage information indicating resource usage related to the distributed processing, and a predetermined threshold value related to the resource usage information;
Referring to the resource usage status information and the predetermined threshold value, when the resource usage status information exceeds the predetermined threshold value, the distributed processing is temporarily stopped, the notification to the maintenance person terminal, and the addition of the resource A resource monitoring function unit that executes at least one of
A processing server comprising: The processing server according to claim 4, wherein the resource usage status information is information on a CPU (Central Processing Unit) or a memory usage rate related to the distributed processing. The processing server according to claim 4, wherein the resource usage status information is information on the number of distributed processes per unit time.   The program for functioning a computer as a processing server as described in any one of Claim 4-6.

Images