JP2022017754A - Container location destination determining program, and container location destination determining method - Google Patents

Abstract

To suppress delay in communication between programs.SOLUTION: The present invention is directed to a container location destination determining program which makes a computer execute processing for specifying, if a resource use rate of a first container located in a first virtual machine exceeds a threshold value, a second program requesting processing to a first program operated in the first container, for specifying a second virtual machine having a second container located therein and having the second program operated therein, and for determining the second virtual machine as a scale-out destination of the first container.SELECTED DRAWING: Figure 16

Description

The present invention relates to a container placement destination determination program and a container placement destination determination method.

With the development of virtualization technology, the cloud that provides various services to users by using virtual machines running in data centers is becoming widespread. In particular, in recent years, a new service called multi-cloud, which combines the clouds of different cloud providers, is becoming widespread. In the multi-cloud, even if a failure occurs in one of the cloud operators, the virtual machines can continue to operate in the cloud of the remaining cloud operators, so that the availability of services can be improved.

The application program for realizing multi-cloud may be executed by the container started in the virtual machine instead of being executed directly by the virtual machine. In the following, the application program is also simply referred to as an application.

A container is a virtual user space that uses only a part of the kernel of the guest OS, and is created by a container engine such as DOCKER (registered trademark). By storing the library that the application refers to at the time of execution in the container in advance, when the application is moved to another virtual machine together with the container, the application can be executed in the same environment as before the movement.

However, there is room for improvement in the multi-cloud using containers in terms of suppressing delays in communication between programs such as applications.

Japanese Unexamined Patent Publication No. 2017-219972

According to one aspect, the purpose is to suppress the delay in communication between programs.

According to one aspect, when the resource utilization of the first container placed in the first virtual machine exceeds the threshold value on the computer, the first program running in the first container is used. The second program requesting processing is specified, the second virtual machine in which the second container in which the second program is running is located, and the second virtual machine is specified. A container placement destination determination program for executing a process, which is determined as a scale-out destination of the first container, is provided.

According to one aspect, delays in communication between programs can be suppressed.

FIG. 1 is a multi-cloud system configuration diagram examined by the inventor of the present application. FIG. 2 is a schematic diagram of each application of the processing request source and the request destination. FIG. 3 is a schematic diagram showing an example of scale-out. FIG. 4 is a schematic diagram showing the distribution of processing to each application when scaling out. FIG. 5 is a schematic diagram showing the definition of the delay time. FIG. 6 is a configuration diagram of an information processing system according to the present embodiment. FIG. 7 is a schematic diagram of the physical server according to the present embodiment. FIG. 8 is a schematic diagram showing each of the virtual machines, containers, and applications operating at each cloud base in the present embodiment. FIG. 9 is a schematic diagram of a system constructed by each of a virtual machine, a container, and an application in the present embodiment. FIG. 10 is a schematic diagram (No. 1) showing a container placement destination determination method according to the present embodiment. FIG. 11 is a schematic diagram (No. 2) showing a container placement destination determination method according to the present embodiment. FIG. 12 is a schematic diagram (No. 3) showing a container placement destination determination method according to the present embodiment. FIG. 13 is a diagram showing a monitoring result of the resource usage rate of each container in the present embodiment. FIG. 14 is a schematic diagram of design information according to this embodiment. FIG. 15 is a schematic diagram of management information according to this embodiment. FIG. 16 is a schematic diagram of the system 30 after scaling out in the present embodiment. FIG. 17 is a functional configuration diagram of the information processing apparatus according to the present embodiment. FIG. 18 is a flowchart of the container placement destination determination method according to the present embodiment. FIG. 19 is a hardware configuration diagram of the information processing apparatus according to the present embodiment.

Prior to the description of the present embodiment, the matters examined by the inventor of the present application will be described.
FIG. 1 is a multi-cloud system configuration diagram examined by the inventor of the present application.

This system 1 is a system for providing services to users, and is constructed by combining a plurality of virtual machine VMs in different regions and businesses that provide clouds.

A region is a region where a data center containing a physical machine running a virtual machine VM resides. For example, "Japan" and "US" are regions. In addition, the business operator is a business operator that provides a virtual machine VM using the data center. In the example of FIG. 1, the system 1 is constructed by the virtual machine VMs provided by each of the business operator A and the business operator B. In the following, the pair of business operator and region will be referred to as a cloud base. For example, (Operator A, Japan), (Operator A, USA), (Operator B, Japan), and (Operator B, USA) are examples of different cloud bases.

When the user of the system 1 operates the data center on-premises, the virtual machine VM may be operated in the data center owned by the user and the virtual machine VM may be incorporated into the system 1.

Further, in this example, a plurality of containers 4 are arranged in each of one virtual machine VM. Then, one application 5 is placed in each container 4. As a result, by moving the container 4 together with the application 5 to another virtual machine VM, it is possible to move the virtual machine VM for each application.

In such a system 1, one service is realized by each of a plurality of applications 5 requesting processing from each other.

FIG. 2 is a schematic diagram of each application 5 of the processing request source and the request destination.
In the example of FIG. 2, it is assumed that each application 5 of "appA", "appB", "appO", "appP", "appW", and "appX" requests the application 5 of "appV" for processing. is doing. When the processing is concentrated on the application 5 of the "appV" in this way, the resources such as the memory and the CPU (Central Processing Unit) allocated to the container 4 of the application 5 of the "appV" are insufficient, and the processing speed of the application 5 becomes high. It may decrease.

To avoid this, create a new virtual machine VM at the cloud base (operator A, USA) where the "appV" app 5 is running, and the "appV" app 5 runs on that virtual machine VM. It is conceivable to scale out the container 4 that is in use. Scale-out is an operation of copying a container 4 running on a certain virtual machine VM to another virtual machine VM together with an application 5 running on the container 4. As a result, since the application 5 of "appV" operates on the two virtual machine VMs of the cloud base (operator A, the United States), it is possible to suppress the concentration of processing on the application 5.

However, when a new virtual machine VM is created in this way, the number of virtual machine VMs included in the system 1 increases, so that the usage fee for the system 1 increases.

Therefore, in order to prevent the usage fee from increasing and to prevent the processing from being concentrated on the application 5 of "appV", the virtual machine VM that already exists in the system 1 should be scaled out as follows. Can be considered.

FIG. 3 is a schematic diagram showing an example of scale-out in this case.
In the example of FIG. 3, the container 4 in which the application 5 of "appV" is operating at the cloud base (operator A, USA) is scaled out to the existing virtual machine VM of the cloud base (business B, USA). .. Here, the name "appV'" obtained by adding the prime to the name "appV" of the application 5 before the scale-out is used as the name of the application 5 at the scale-out destination.

FIG. 4 is a schematic diagram showing the distribution of processing to each application 5 when scaled out in this way.

In the example of FIG. 4, the load balancer 6 is started in an appropriate virtual machine VM, and the load balancer 6 is used to evenly distribute the processing to each application 5 of "appV" and "appV'". As a result, it is possible to suppress a shortage of resources such as memory and CPU allocated to the container 4 of the application 5 of the "appV". Moreover, since the scale-out destination virtual machine VM originally exists in the system 1, it is not necessary to newly create the virtual machine VM, and it is possible to prevent the usage fee of the system 1 from increasing.

However, if the cloud base of the scale-out destination virtual machine VM is different from the cloud base of the scale-out source virtual machine VM as in this example, there is a high possibility that the delay time between the applications 5 will increase.

FIG. 5 is a schematic diagram showing the definition of the delay time.
Here, the time from when the application 5 of "appA" requests the application 5 of " _appB " to process and the result is notified is defined as the response time TR. Further, the time required for the application 5 of "appB" to perform the processing is defined as the processing time T _P. In this case, the delay time T _L is defined by T _{R --TP} . According to this definition, the delay time T _L is the length of the delay due to the communication path. Therefore, the delay time T _L tends to increase between the geographically distant apps 5, such as between the apps 5 in different cloud bases.

The requirement of the system 1 (see FIG. 3) that provides the service to the user may be restricted that the delay time T _L should be equal to or less than the upper limit value (for example, 5 msec). At this time, if the two apps 5 are placed in different cloud bases due to scale-out as described above, for example, the delay time T _L between these apps 5 becomes 10 msec, and the requirements of the system 1 cannot be satisfied. Will end up.

(The present embodiment)
FIG. 6 is a configuration diagram of an information processing system according to the present embodiment.
The information processing system 20 is a system that provides services to users by a multi-cloud, and has a plurality of physical servers 22 and an information processing device 23 that are interconnected via a network 21 such as the Internet.

Of these, the physical server 22 is a computer housed in each of the data centers of the business operator A and the business operator B that provide the cloud. Here, it is assumed that operators A and B have data centers in different regions, "Japan" and "US", respectively. The information processing system 20 may include the physical server 22 owned by the user who receives the service.

A virtual machine is started on each physical server 22, and a container is started inside the virtual machine. Then, each container executes various applications for providing a service to the user.

The information processing device 23 is a PC (Personal Computer), a server, or the like for controlling to scale out the container to another virtual machine when the resource usage rate of the container running in a certain virtual machine increases. It is a computer.

FIG. 7 is a schematic diagram of the physical server 22.
As shown in FIG. 7, the physical server 22 has a physical resource 25 such as a CPU (Central Processing Unit) and a memory, and the resource 25 executes the host OS 26.

Further, the physical server 22 starts a plurality of virtual machines 27 on the host OS 26. In the following, each virtual machine 27 is identified by a character string such as "VM ₁ ", "VM ₂ ", and so on. Each virtual machine 27 uses a portion of the resource 25 to start the container 28. The container engine used by each virtual machine 27 to start the container 28 is, for example, DOCKER (registered trademark). Then, each container 28 executes one application 29.

FIG. 8 is a schematic diagram showing each of the virtual machine 27, the container 28, and the application 29 operating at each cloud base.

As shown in FIG. 8, for example, a virtual machine 27 of "VM ₁₀ " is operating in a cloud base (operator A, USA), and "appV", "appW", and "appW" are operated in each container 28 of the virtual machine 27. Each app 29 of "appX" is operating.

FIG. 9 is a schematic diagram of the system 30 constructed by each of the virtual machine 27, the container 28, and the application 29.

The system 30 is a system used by the user and has a plurality of virtual machines 27 connected to each other by the network 21. In the system 30, if the processing is concentrated on a certain application 29, the resource 25 (see FIG. 7) allocated to the container 28 executing the application 29 may be insufficient. In this case, the information processing apparatus 23 (see FIG. 6) scales out the container 28 as follows.

10 to 12 are schematic views showing a container placement destination determination method according to the present embodiment.

First, as shown in FIG. 10, the information processing apparatus 23 identifies the container 28 in which the application 29 of "appV" is operating as the container 28 to be scaled out. As an example, the information processing apparatus 23 monitors the resource usage rate of each container 28, and sets the container 28 in which the application 29 of "appV" whose resource usage rate exceeds the threshold value is operating as the container 28 that needs to be scaled out. Identify.

FIG. 13 is a diagram showing a monitoring result of the resource usage rate of each container 28.
In FIG. 13, each container 28 is distinguished by the name of the application 29 running on each container 28. Further, in this example, it is assumed that each of the CPU usage rate and the memory usage rate is adopted as the resource usage rate of each container 28, and 85% is adopted as the threshold value of the resource usage rate. In this case, the information processing apparatus 23 specifies the container 28 in which either the CPU usage rate or the memory usage rate exceeds 95% as the container 28 that needs to be scaled out.

See FIG. 10 again.
Next, the information processing device 23 identifies the requesting application 29 requesting the application 29 of the "appV" to perform processing. The method of specifying the application 29 is not particularly limited. Here, the information processing device 23 identifies the requesting application 29 by referring to the design information associated with each application 29.

FIG. 14 is a schematic diagram of the design information 40.
The design information 40 is information generated when the developer designs the information processing system 20, and is information in which the application 29 of the request source and the application 29 of the request destination are associated with each other. Here, the symbol "○" is added when the two applications 29 have a relationship between the request source and the request destination for processing, and the symbol "×" is added when the two applications 29 do not have a relationship between the request source and the request destination. There is. According to FIG. 14, the information processing apparatus 23 has "appA", "appB", "appO", "appP", as the requesting application 29 requesting processing from the "appV" application 29. Each application 29 of "appW" and "appX" is specified.

See FIG. 10 again.
Next, the information processing device 23 identifies a virtual machine 27 in which each application 29 requesting processing from the application 29 of "appV" is arranged. In the example of FIG. 10, the information processing apparatus 23 identifies the virtual machines 27 of "VM ₁ ", "VM ₇ ", and "VM ₁₀ ".

The method for specifying the virtual machine 27 is not particularly limited. For example, the information processing apparatus 23 generates management information by executing a container management program such as Kubernetes, and identifies the virtual machine 27 based on the management information.

FIG. 15 is a schematic diagram of the management information 41.
The management information 41 is information in which the application 29 and the virtual machine 27 that executes the application 29 are associated with each other. In the example of FIG. 15, the virtual machine 27 of "VM ₁ " executes each container 28 of "appA" and "appB", and the virtual machine 27 of "VM ₇ " executes each container 28 of "appO" and "appP". Is running. Further, the virtual machine 27 of "VM ₁₀ " is executing each container 28 of "appW" and "appX".

Next, as shown in FIG. 11, the information processing apparatus 23 uses the virtual machines 27 of "VM ₁ " and "VM ₇ " specified above in the container 28 running the application 29 of "appV". Decide as a scale-out destination. Since the application 29 of "appV" is already executed in the virtual machine 27 of "VM ₁₀ ", the information processing apparatus 23 excludes the virtual machine 27 of "VM ₁₀ " from the scale-out destination.

After that, the information processing apparatus 23 scales out the container 28 running the application 29 of the “appV” to each of the virtual machines 27 of the “VM ₁ ” and the “VM ₇ ”. Here, the name of the application 29 executed by the container 28 scaled out to the virtual machine 27 of "VM ₁ " is "appV'", and the application 29 executed by the container 28 scaled out to the virtual machine 27 of "VM ₇ " is executed. The name of is "appV".

Next, as shown in FIG. 12, the information processing apparatus 23 requests the processing of each of the applications 29 of "appA" and "appB" to the virtual machine 27 of "VM ₁ " which is the same as those of these applications 29. Change to app 29 of "appV'" in. Similarly, in the information processing device 23, the request destination to which each application 29 of "appO" and "appP" requests processing is the "appV" in the virtual machine 27 of the same "VM ₇ " as these applications 29. Change to app 29.

This completes the basic processing of the container placement destination determination method according to the present embodiment.

FIG. 16 is a schematic diagram of the system 30 after scaling out as described above.

As shown in FIG. 16, after scaling out, the request destination to which each application 29 of "appA" and "appB" requests processing operates on the same "VM ₁ " virtual machine 27 as these applications 29. It becomes the application 29 of "appV'". Similarly, the request destination to which each application 29 of "appO" and "appP" requests processing is the application 29 of "appV" running on the virtual machine 27 of the same "VM ₇ " as these applications 29.

As a result, it is possible to suppress the generation of processing requests between the applications 29 in different cloud bases due to scale-out, and it is possible to suppress the delay in communication between the applications 29. As a result, if the requirements of the system 30 specify an upper limit of the delay time, the requirements can be satisfied even after the scale-out.

Next, the functional configuration of the information processing apparatus 23 according to the present embodiment will be described.

FIG. 17 is a functional configuration diagram of the information processing apparatus 23 according to the present embodiment.
As shown in FIG. 17, the information processing apparatus 23 includes a storage unit 51, a communication unit 52, and a control unit 53.

Of these, the storage unit 51 stores the above-mentioned design information 40 (FIG. 14) and management information 41 (see FIG. 15).

Further, the communication unit 52 is a communication interface for connecting the information processing device 23 to the network 21 (see FIG. 6).

The control unit 53 is a processing unit that controls each unit of the information processing device 23, and is a resource monitoring unit 55, an application identification unit 56, a virtual machine identification unit 57, a determination unit 58, a scale-out execution unit 59, and a request destination change. It has a part 60.

Of these, the resource monitoring unit 55 is a processing unit that monitors the resource usage rates of all the containers 28 included in the system 30 (see FIG. 9). The resource usage rate to be monitored includes, for example, the memory usage rate and the CPU usage rate.

The application specifying unit 56 is a processing unit that identifies the application 29 operating in the container 28 whose resource usage rate exceeds the threshold value, and further specifies the requesting application 29 requesting the application 29 to process. .. As an example, the application specifying unit 56 identifies the requesting application 29 with reference to the design information 40 in FIG. In the example of FIG. 14, when the resource usage rate of the container 28 in which the application 29 of "appV" is operating exceeds the threshold value, the application specifying unit 56 performs "appA", "appB", "appO", and "appO". Each application 29 of "appP", "appW", and "appX" is specified as a request source.

The virtual machine specifying unit 57 is a processing unit that identifies the virtual machine 27 in which the container 28 in which the application 29 specified by the application specifying unit 56 is operating is arranged by referring to the management information 41 in FIG. .. In the example of FIG. 15, the virtual machine specifying unit 57 identifies each virtual machine 27 of "VM ₁ ", "VM ₇ ", and "VM ₁₀ ".

The determination unit 58 is a processing unit that determines the virtual machine 27 specified by the virtual machine identification unit 57 as the scale-out destination of the container 28 whose resource usage rate exceeds the threshold value. In the case of FIG. 15, the determination unit 58 determines each of the virtual machines 27 of "VM ₁ " and "VM ₇ " as the scale-out destination of the container 28 running the application 29 of "appV". The decision unit 58 excludes the virtual machine 27 of the "VM ₁₀ " that has already executed the application 29 of the "appV" from the scale-out destination.

As shown in FIG. 11, the scale-out execution unit 59 is a processing unit that actually scales out the container 28 to the scale-out destination virtual machine 27 determined by the determination unit 58.

The request destination changing unit 60 is a processing unit that changes the request destination to which the application 29 specified by the application specifying unit 56 requests processing to the application 29 that operates in the container 28 after the scale-out. In the example of FIG. 12, the request destination changing unit 60 changes the request destination to which each application 29 of "appA" and "appB" requests processing to the application 29 of "appV'". Similarly, the request destination changing unit 60 changes the request destination to which each application 29 of "appO" and "appP" requests processing to the application 29 of "appV".

Next, a method for determining the container placement destination according to the present embodiment will be described.
FIG. 18 is a flowchart of the container placement destination determination method according to the present embodiment.

First, the resource monitoring unit 55 monitors the resource usage rates of all the containers 28 included in the system 30 (see FIG. 9) (step S11).

Next, the application specifying unit 56 identifies the application 29 operating in the container 28 whose resource usage rate exceeds the threshold value, and further specifies the requesting application 29 requesting the application 29 to process (the application specifying unit 56). Step S12).

Next, the virtual machine specifying unit 57 identifies the virtual machine 27 in which the container 28 in which the requesting application 29 is operating is arranged (step S13).

Subsequently, the determination unit 58 determines the virtual machine 27 specified in step S13 as the scale-out destination of the container 28 whose resource usage rate exceeds the threshold value (step S14).

Next, the scale-out execution unit 59 scales out the container 28 whose resource usage rate exceeds the threshold value to the scale-out destination virtual machine 27 determined in step S14 (step S15).

Then, the request destination change unit 60 changes the request destination to which the application 29 specified in step S12 requests processing to the application 29 that operates in the container 28 after the scale-out (step S16).

This completes the basic processing of the container placement destination determination method according to the present embodiment.

According to the above-described embodiment, the container 28 in which the requesting application 29 requesting the application 29 to process the scale-out destination of the container 28 whose resource usage rate exceeds the threshold value is arranged. Let's assume that the virtual machine 27 is running. As a result, since the two applications 29 of the processing request source and the request destination operate on the same virtual machine 27 after the scale-out, the processing request does not occur between the two applications 29 in different cloud bases. As a result, the delay time T _L between the applications 29 can be suppressed, and the possibility that the delay time T _L does not meet the requirements of the system 30 can be reduced.

(Hardware configuration)
FIG. 19 is a hardware configuration diagram of the information processing apparatus 23.
As shown in FIG. 19, the information processing device 23 includes a storage device 23a, a memory 23b, a processor 23c, a communication interface 23d, a display device 23e, an input device 23f, and a medium reading device 23g. Each of these parts is connected to each other by a bus 23i.

Of these, the storage device 23a is a non-volatile storage such as an HDD (Hard Disk Drive) or SSD (Solid State Drive), and stores the container placement destination determination program 100 according to the present embodiment.

The container placement destination determination program 100 may be recorded on a recording medium 23h that can be read by a computer, and the processor 23c may be made to read the container placement destination determination program 100 via the medium reading device 23g.

Examples of such a recording medium 23h include a physically portable recording medium such as a CD-ROM (Compact Disc-Read Only Memory), a DVD (Digital Versatile Disc), and a USB (Universal Serial Bus) memory. Further, a semiconductor memory such as a flash memory or a hard disk drive may be used as the recording medium 23h. These recording media 23h are not temporary media such as a carrier wave having no physical form.

Further, the container placement destination determination program 100 may be stored in a device connected to a public line, the Internet, a LAN, or the like. In that case, the processor 23c may read and execute the container placement destination determination program 100.

On the other hand, the memory 23b is hardware that temporarily stores data such as DRAM (Dynamic Random Access Memory), and the container placement destination determination program 100 is deployed on the hardware.

The processor 23c is hardware such as a CPU and a GPU (Graphical Processing Unit) that control each part of the information processing device 23. Further, the processor 23c executes the container placement destination determination program 100 in cooperation with the memory 23b.

By executing the container placement destination determination program 100 in cooperation with the memory 23b and the processor 23c in this way, the control unit 53 (see FIG. 17) of the information processing apparatus 23 is realized. The control unit 53 includes a resource monitoring unit 55, an application identification unit 56, a virtual machine identification unit 57, a determination unit 58, a scale-out execution unit 59, and a request destination change unit 60.

Further, the storage unit 51 (see FIG. 17) is realized by the storage device 23a and the memory 23b.

Further, the communication interface 23d is hardware such as a NIC (Network Interface Card) for connecting the information processing device 23 to the network 21 (see FIG. 6). The communication unit 52 (see FIG. 17) is realized by the communication interface 23d.

The display device 23e is hardware such as a liquid crystal display and a touch panel for displaying various information.

Further, the input device 23f is hardware such as a keyboard and a mouse for the administrator of the system 30 to input various data to the information processing device 23.

The medium reading device 23g is hardware such as a CD drive, a DVD drive, and a USB interface for reading the recording medium 23h.

1 ... system, 4 ... container, 5 ... app, 6 ... load balancer, 20 ... information system, 21 ... network, 22 ... physical server, 23 ... information processing device, 25 ... resource, 27 ... virtual machine, 28 ... container , 29 ... App, 30 ... System, 40 ... Design information, 41 ... Management information, 51 ... Storage unit, 52 ... Communication unit, 53 ... Control unit, 55 ... Resource monitoring unit, 56 ... App identification unit, 57 ... Virtual machine Specific part, 58 ... decision part, 59 ... scale-out execution part, 60 ... request destination change part.

Claims (5)

On the computer
When the resource usage rate of the first container placed in the first virtual machine exceeds the threshold value, the first program running in the first container is requested to process. Identify the program of 2
Identify the second virtual machine where the second container running the second program is located.
Determining the second virtual machine as the scale-out destination for the first container.
A container placement destination determination program for executing processing. To the computer
Scale out the first container to the second virtual machine and
The request destination to which the second program requests processing is changed to the first program operating in the first container placed in the second virtual machine by scale-out.
The container placement destination determination program according to claim 1, for executing the process. The container placement destination determination program according to claim 1, wherein the first virtual machine and the second virtual machine are placed in different cloud bases. To the computer
By referring to the storage unit that stores the information in which the first program and the second program are associated with each other, the second program for which processing is requested to the first program is specified. ,
The container placement destination determination program according to claim 1, for executing the process. The computer
When the resource usage rate of the first container placed in the first virtual machine exceeds the threshold value, the first program running in the first container is requested to process. Identify the program of 2
Identify the second virtual machine where the second container running the second program is located.
Determining the second virtual machine as the scale-out destination for the first container.
A method for determining a container placement destination, which is characterized by executing a process.

Images