JP6239400B2 - Control device - Google Patents

Description

  The present invention relates to a virtualized control server management method, control server management apparatus, and program, and in particular, a virtual control server management method for allocating computer resources in accordance with the control period and control processing time of each control virtual server, The present invention relates to a science device and a program.

  In systems using server virtualization technology such as cloud computing, in order to improve the utilization efficiency of server resources such as memory and processor, and to maintain the service level against the load that varies with time and time, There exists a technique described in 2011-258119 gazette (patent document 1). This gazette states that “a method of controlling a cluster composed of a plurality of virtual machines, which calculates a predicted value of load information after a predetermined time for the cluster and calculates the current computer resources of the virtual machines constituting the cluster. The allocation amount and the allocation amount of the computer resource after a predetermined time are held, the computer resource changing means that can be used to change the cluster configuration is held, and the current computer assigned to the virtual computer constituting the cluster The combination of changing means is selected from the resource allocation amount and the predicted value of load information, the scheduled execution time of the changing means is set, and the changing means that has reached the scheduled execution time is notified to the virtualization unit and executed. Is disclosed.

JP2011-258119

  In the control system, the control server periodically performs real-time processing based on the control cycle. In Patent Document 1, the processor usage rate is monitored for the allocation of computer resources, particularly the allocation of processor resources, but the computer resource allocation is performed from the viewpoint of protecting the real-time property in the periodic operation of the control server. Absent. Therefore, when applying the technique of Patent Document 1 to a control server virtualization apparatus, it is difficult to maintain a service level for real-time control. For example, when a plurality of virtual control servers operate on the same CPU core at the same time, the same CPU core is alternately used by a plurality of virtual control servers, so the operation time increases and the real-time property is impaired.

  Therefore, an object of the present invention is to provide a technique for maintaining periodic real-time control when a control system is operated on a plurality of host servers using a plurality of virtual control servers.

  In order to solve the above problems, one of the representative control devices of the present invention is to add a task received from a user when a task is newly added to a virtual control server in which a control application operates in a server virtualization environment. A virtual control server schedule table for storing the execution cycle and the maximum execution time is provided. The hypervisor refers to the virtual control server schedule table, checks the control cycle and maximum processing time of the already operating virtual control server, and the newly added virtual control server has the same cyclic operation as another virtual server on the same processor. A virtual control server additional control mechanism that efficiently aggregates physical servers so that they do not overlap with each other. The virtual control server relocation mechanism performs defragmentation processing of the virtual control server so that the CPU free time can be long in the host server and between multiple host servers, either automatically at regular intervals or according to instructions from the user. carry out.

  According to the present invention, it is possible to easily execute a periodically executed program without impairing real-time performance.

The block diagram which shows the structure of the control apparatus which consists of a some host server which concerns on the Example of this invention, a some virtual control server, and a control object apparatus. The figure which shows the structure of the virtual control server schedule table in an Example. The figure which shows the virtual control server schedule operation | movement in a host server based on the information of the virtual control server schedule table in an Example. The flowchart which shows embodiment of the process sequence which adds an Example virtual control server newly. The figure which shows the structure of the virtual control server schedule table when virtual control server "D" and "E" are newly added to the virtual control server schedule table in an Example. The figure which shows the virtual control server schedule operation | movement in a host server based on the information of the virtual control server schedule table in an Example. The figure which shows the structure of the virtual control server schedule table when virtual control server "F" and "G" are added to the virtual control server schedule table in an Example. The figure which shows the virtual control server schedule operation | movement in a host server based on the information of the virtual control server schedule table in an Example. The figure which shows the structure of the virtual control server schedule table after performing a virtual control server defragmentation process in the virtual control server schedule table in an Example. The figure which shows the virtual control server schedule operation | movement in a host server based on the information of the virtual control server schedule table in an Example. The flowchart which shows the virtual control server defragmentation process procedure in an Example. The figure which shows the operation example of the virtual control server defragmentation process between the host servers in an Example. The figure which shows the structure of the virtual control server schedule table which added the item of "additional information" to the virtual control server schedule table in an Example. The figure which shows the virtual control server schedule operation | movement in a host server based on the information of the virtual control server schedule table in an Example.

  Hereinafter, examples will be described with reference to the drawings.

  FIG. 1 is a block diagram illustrating a configuration of a control device according to the embodiment. The control device 10 includes a plurality of host servers 20. Each host server includes a physical server 30. The physical server 30 includes a memory 32 for operating a program and a CPU core 34. A plurality of CPU cores 34 may be provided. The host server 20 includes a hypervisor 40 that operates on the physical server 30.

  The hypervisor 40 includes a virtual control server schedule table 45 for scheduling a plurality of virtual control servers 50. The hypervisor 40 determines the CPU core 34 for operating the virtual control server and the operation timing based on the schedule information stored in the virtual control server schedule table 45. In addition, the hypervisor 40 includes a virtual control server addition control mechanism 46 that performs a virtual control server addition process, and a virtual control server relocation mechanism that performs a virtual control server defragmentation process.

  The virtual control server 50 includes a control application 52 and a task scheduler 54. The control application 52 controls the control target device 60. The task scheduler 54 schedules the execution of the control application 52 in cooperation with the hypervisor 40. The control device 10 is connected to a plurality of control target devices 60 via the communication device 70. A plurality of virtual control servers operating on the plurality of host servers 20 operate periodically to control the control target device 60.

  FIG. 2 shows a specific example of the virtual control server schedule table 45 in the embodiment. The virtual control server schedule table includes items of “virtual control server name”, “state”, “control cycle”, “maximum execution time”, “operating CPU core number”, and “period startup time difference”. “Status” indicates the status of the virtual control server, and can be a value of “operation” or “stop” (including temporary stop). “Control cycle” indicates the cycle at which the virtual control server is activated. “Maximum execution time” indicates the maximum execution time after the virtual control server is activated.

  Once activated, the virtual control server continues the process until the periodic process execution is completed. In other words, when the task scheduler in the virtual control server communicates with the hypervisor and the control application in the virtual control server is executing cyclic processing, the execution order is set so that execution does not move to other virtual control servers. Control.

  When the periodic processing of the control application in the virtual control server ends, the task scheduler communicates with the hypervisor and notifies the other virtual control server that control may be passed. The “control cycle” and “maximum execution time” are determined from the control contents of the virtual control server, and are input by the user.

  “Operation CPU core number” indicates the CPU core of the physical server on which the virtual control server operates. The “periodic activation time difference” indicates a time difference at which the virtual control server is activated. The period activation time difference is a time difference in the least common multiple period of the control period of each virtual control server. The “operating CPU core number” and “period startup time difference” are calculated by the hypervisor virtual control server additional control mechanism when the virtual control server starts up, based on the “control cycle” and “maximum execution time” input from the user. , To decide. Details of the calculation method will be described in “Additional processing of virtual control server” (FIG. 4).

  FIG. 3 is a diagram illustrating the activation / operation timing of the virtual control server in the CPU core in the host server in the embodiment. Here, it is assumed that the host server includes two CPU cores (CPU core 1 and CPU core 2).

  FIG. 3 corresponds to the virtual control server schedule table of FIG. That is, the virtual control servers “A”, “B”, and “C” are operating with their respective parameters. You can see the startup and operation time of the virtual control server in the period of 6 seconds, which is the least common multiple of the control period of the virtual control server. The reason for using the least common multiple is that each virtual control server operates periodically, and therefore all virtual control servers repeat the same operation pattern in the least common multiple cycle.

  In this example, the virtual control server A has a control cycle of 1 second, a maximum execution time of 200 milliseconds, an operating CPU core number of 1, and a period startup time difference of 0 milliseconds. It operates for 200 milliseconds, 1 second to 1200 milliseconds, 2 seconds to 2200 milliseconds, 3 seconds to 3200 milliseconds, 4 seconds to 4200 milliseconds, 5 seconds to 5200 milliseconds. The same applies to the virtual control servers B and C.

  In this example, the periodic operation of the virtual server is described as an example, but in a system without a virtual server, it can be replaced with a task or the like.

  FIG. 4 is a flowchart showing an addition processing procedure of the virtual control server addition control mechanism. The operation based on the flowchart is as follows.

  Step 100: Accept input from the user of the virtual control server name, control cycle, and maximum execution time of the virtual control server to be added, and store them in the virtual control server schedule table.

  Step 101: Calculate the least common multiple of the control periods of all virtual control servers from the virtual control server schedule table.

  Step 102: Step 104 is repeated for all CPU cores (C) provided in the host server in ascending order of CPU core numbers.

  Step 104: The CPU core (C) is operated without overlapping the operation time of the other virtual control server and the virtual control server to be added within the periodic operation having the least common multiple of the control cycle calculated in Step 101 as a cycle. Determine whether you can. Judgment is made when the virtual control server cycle to be added is T, the maximum operation time is ΔT, the least common multiple of the control cycle is L, the startup time difference is W, and the maximum integer less than L / T is N. [W, W + ΔT], [W + T, W + T + ΔT], [W + 2 * T, W + 2 * T + ΔT], ..., [W + N * T, W + N * T + ΔT] is performed by determining whether it overlaps with other virtual control servers.

  Step 105: It is determined whether there is a physical CPU core that has not yet performed the process of Step 104. If there is, return to 102, and if not, execute 106.

  Step 106: If it is impossible to operate all physical CPU cores without overlapping the operation time with other virtual control servers in step 104, an error message indicating that no more virtual control servers can be added is displayed. Output. The error message includes the name of the virtual control server to be added and the host server name.

  Step 108: If the operation can be performed without overlapping the operation time with other virtual control servers in Step 104, the smallest one of the settable start time differences W is selected and combined with the operation CPU core (C). Store in the virtual control server schedule table.

  Step 110: Start the virtual control server in accordance with the start time difference recorded in the virtual control server schedule table.

  By the above virtual control server addition process, the virtual control servers can be operated without overlapping the operation time and with each control cycle and execution time secured. Since the virtual control server is not subject to interference from other virtual control servers, real-time performance can be ensured.

Hereinafter, examples of the present invention will be described with specific examples of the virtual control server addition processing.
FIG. 2 shows a virtual control server schedule table when virtual control servers “A”, “B”, and “C” are added in order. When adding the virtual control server “A” for the first time, since there is no virtual control server that is already running, the least common multiple of the control cycle of the virtual control server is the same as the control cycle of the virtual control server “A”, that is, In this case, it is 1 second.

  Since there is no other virtual control server in the cyclic activation time difference, 0 seconds is set and “A” is activated. When the virtual control server “B” is added in addition to “A”, the least common multiple of the control cycles of all virtual control servers is 2 seconds. In the 2-second period, the virtual control server “A” is operating during the interval 0 second to 0.2 second and 1 second to 1.2 seconds. 2 seconds is set and “B” is activated. Furthermore, when adding the virtual control server “C”, the least common multiple of the control cycle of the virtual control server is 6 seconds. The activation time difference of “C” is set to 0.4 seconds so that the operation time does not overlap with “A” and “B”, and activation is performed.

  The virtual machine scheduling operation in the host server when the virtual control servers “A”, “B”, and “C” are sequentially added is as shown in FIG.

  FIG. 5 shows that in the virtual control server addition process (FIG. 4), in addition to “A”, “B”, “C” virtual control servers, “D” and “E” virtual control servers are added in order. The virtual control server schedule table at the time of doing is shown. When adding the virtual control server “D”, in order to operate the CPU core 1 so that it does not overlap with the already running virtual control server in the least common multiple cycle of the control cycle, Must be 1.2 seconds.

  With other values of the cyclic activation time difference, there are cases where the operation overlaps with the already activated virtual control server. Subsequently, when “E” is added, even if the CPU core 1 tries to operate “E”, there is no periodic activation time difference that can be operated without overlapping with other virtual control servers. For this reason, “E” tries to operate in the CPU core 2. In CPU core 2, before “E” is activated, there is no virtual control server that has already been activated. Therefore, “E” can be activated by setting 0 seconds as the periodical activation time difference. Here, it is assumed that the least common multiple of the control period is shared between the CPU cores 1 and 2.

  FIG. 6 shows a virtual machine scheduling operation in the host server when virtual control servers “D” and “E” are added in addition to the situation in FIG. The virtual control server “D” has a period of 1.2 to 1.4 seconds, 3.2 to 3.4 seconds, and 5.2 to 5.4 seconds in the 6-second cycle in CPU core 1. Operate. The virtual control server “E” operates in the CPU core 2 for 0 to 0.2 seconds, 2 to 2.2 seconds, and 4 to 4.2 seconds.

  FIG. 7 shows a virtual control server schedule table when the virtual control server is repeatedly operated and stopped. The virtual control server stop process will be described. When stopping a virtual control server that is running, stop the operation of the virtual control server, and then select "-" for the operating CPU core number of the virtual control server to be stopped and the cyclic start time difference in the virtual control server schedule table. Record. The symbol “-” represents undefined.

  The virtual control server defragmentation process will be described. The virtual control server defragmentation process is a process of performing defragmentation when the operation section of the virtual control server is in a fragmentation state in the least common multiple cycle of the control cycle when the virtual control server is repeatedly activated and stopped. By performing defragmentation, it becomes easy to secure a long-lasting operation section, so a new virtual control server can be added to effectively use the free CPU time.

  FIG. 8 shows a state of the schedule of the virtual control server in the host server based on the virtual control server schedule table of FIG. As can be seen from FIG. 8, fragmentation of free CPU time occurs. In such a case, even if a new virtual control server with a control period of 3 seconds and a maximum execution time of 0.7 seconds is added to the host server, it cannot be added because there is no continuous free CPU time.

  In such a situation, the virtual control server defragmentation process is a process of collecting virtual control servers operating in the CPU core 2 in the CPU core 1 and ensuring continuous free CPU time in the CPU core 2.

  FIG. 9 shows a state in which a virtual control server “H” having a control cycle of 3 seconds and a maximum execution time of 0.7 seconds is further added after performing the virtual control server defragmentation processing from the situation of FIG. 7 and FIG. This is a virtual control server schedule table.

  FIG. 10 shows a virtual control server schedule on the host server based on the virtual control server schedule table of FIG. The virtual control server “H” can be allocated to the CPU core 2 by moving the virtual control server operating on the CPU core 2 to the CPU core 1 by the virtual control server defragmentation process.

  FIG. 11 is a flowchart when the virtual control server relocation mechanism 47 performs virtual control server defragmentation processing. The virtual control server defragmentation process will be described in detail based on FIG.

  Step 200: Calculate the least common multiple of the control periods of all virtual control servers from the virtual control server schedule table.

  Step 202: Steps 204 and 206 are repeated for all physical CPU cores (C) in the host server.

  Step 204: Step 206 is repeated for all virtual servers (V) operating on the physical CPU core (C).

  Step 206: From the CPU cores having the smallest CPU number among the CPU cores having a smaller CPU number than the physical CPU core (C), the virtual control server on the CPU core without overlapping the operation time with other virtual control servers. It is determined whether (V) can be operated without changing the period startup time difference. If it can be operated, step 208 is executed. If the operation is impossible, step 204 is repeated.

  Step 208: Perform a migration process between physical CPU cores of the virtual control server. In the migration process between physical CPU cores, the virtual control server is migrated in the same manner as the task migration mechanism between CPU cores performed by a general multi-core multitask OS.

  Step 210: Rewrite the operation CPU core item in the virtual control server schedule table from the migration source CPU number to the migration destination CPU number. Thereafter, the migrated virtual control server operates on the migration destination CPU.

  Step 211: It is determined whether there is a virtual control server that has not yet performed the process of Step 206. If there is, the process returns to Step 204, and if not, Step 212 is performed.

  Step 212: It is determined whether there is a physical CPU core that has not yet performed the process of Step 206. If there is, the process returns to Step 202, and if not, the process ends.

  A specific example of the virtual machine migration process will be described with reference to FIG. FIG. 9 shows a situation where the virtual control servers “F” and “G” are migrated from the CPU core 2 to the CPU core 1 and then the virtual control server “H” is added from the situation of FIG. . In the situation of FIG. 7, when the virtual control server migration process is started, step 206 is performed for the virtual control servers “A”, “B”, “D”, and “E”. It cannot be operated without overlapping the operation time. When step 206 is performed for “F”, the virtual CPU “F” can be operated in the physical CPU core 1 without changing the periodical activation time difference. In other words, the virtual control server “F” operates in the CPU core 2 in the interval of 0.4 to 0.7 seconds, 3.4 to 3.7 seconds, among the least common multiple of 6 seconds of the control cycle. However, since there is no virtual control server operating in the section in the CPU core 1, it is determined that the virtual control server “F” can be migrated from the CPU core 2 to the CPU core 1. After the determination, migration is performed. Thereafter, the virtual control server “F” operates on the CPU core 1. Similarly, the virtual control server “G” is also migrated from the CPU core 2 to the CPU core 1, and the virtual control server defragmentation process is terminated. After the flag processing in the virtual control server, the CPU free time has increased in CPU core 2, so a new virtual control server “H” with a control cycle of 3 seconds and a maximum execution time of 0.7 seconds is added. be able to.

  The virtual control server migration process may be performed periodically by an instruction from a user or automatically when the control device is operating.

  The virtual control server migration process within the same host server has been described above. However, there is an embodiment in which the migration process is performed between host servers.

  FIG. 12 shows how the virtual control server is migrated between the host servers by the host server 2 and the migration process between the host servers 3 to the host server 1. The inter-host server migration process is executed automatically according to an instruction from the user or periodically. In this example, the virtual control server “F” operating on the CPU core 2 of the host server 2 is migrated to the CPU core 2 of the host server 1 between the host servers. Further, from the host server 3, the virtual control server “G” operating on the CPU core 2 is migrated between the host servers to the CPU core 2 of the host server 1. The host-server migration may be performed while the virtual control server is operating or in a stopped state. When performing migration between host servers while operating the virtual control server, it is performed by live migration of the virtual server, which is a technique generally known in a virtual environment. As described above, by performing the migration process between the host servers, it is possible to consolidate the virtual control servers in the migration migration destination, and to make room for CPU time in the migration source host server.

  In the present embodiment, a column “additional information” is newly added to the virtual control server schedule table 45 handled in the first embodiment.

  13 and 14, the virtual control server addition process of the virtual control server addition control mechanism 46 will be described below only with respect to the changes from the first embodiment.

  In FIG. 13, “control target area” is added as “additional information” to the virtual control server schedule table of the first embodiment. The control target area represents an area division where the control target exists, which is controlled by the corresponding virtual control server. By giving the control target area as additional information, a temporary load is generated on the virtual control server that controls the same area, and the virtual control server is executed beyond the set maximum execution time. It can be provided when a high load is applied.

The additional information is input by the user when a new virtual control server is added. FIG. 13 shows a state when virtual control servers “A”, “B”, “C”, and “D” are added in order. Here, when adding the virtual control server “B” after adding the virtual control server “A”, the control target area that is additional information is the same, so it operates continuously with the virtual control server of “A”. As a margin value, the maximum execution time of “A” is multiplied by 1.5, and “A” is operating from 0 to 0.6 seconds, and from 3 to 3.6 seconds. Assuming that the period start time difference of “B” is 0.6 seconds.
It is possible to prevent “A” and “B” having the same additional information from operating continuously and spreading the processing delay. As a result, even if the load is concentrated on the virtual control server that controls the region (a), the margin is set in advance and the operation time is set, so if the virtual control server exceeds the maximum execution time Even if it exists, there exists an effect that it is hard to exert influence on the virtual control server which controls the same area.

  FIG. 14 shows a virtual control server schedule on the host server based on the virtual control server schedule table shown in FIG. From this figure, it can be understood that virtual control servers in the same control target area are not operating continuously.

  Regarding the virtual control server addition process of the virtual control server addition control mechanism 46, the method of using the additional information to be given to the virtual control server has been described, but the same concept applies to the use of the additional information in the virtual control server defragmentation process. Is done. That is, in the defragmentation process, the defragmentation process is performed so that virtual control servers having the same additional information do not operate continuously. The same applies to the defragmentation process across the host servers.

  As described above, the method of using the additional information in the virtual control server schedule table has been described. However, the additional information can be added to any attribute that can be added to the virtual control server as well as the control target area. For example, it includes the type of control information handled by the virtual control server, the importance of the control target, the load pattern applied to the virtual control server, and the like.

  For example, when the present invention is applied to a railway operation system, it may be possible to specify a railway route or line section as additional information.

  In this case, if the processing provided by each virtual control server for each route is not performed correctly, the route cannot be operated, so by assigning it to an independent CPU core for each route or line unit, It is possible to prevent the processing delay of the virtual control server trying to process the route or line section from spreading to the processing of another route or line section.

  In addition, this invention is not limited to an above-described Example, Various modifications are included. For example, the above-described embodiments have been described in detail for easy understanding of the present invention, and are not necessarily limited to those having all the configurations described. Further, a part of the configuration of one embodiment can be replaced with the configuration of another embodiment, and the configuration of another embodiment can be added to the configuration of one embodiment. Further, it is possible to add, delete, and replace other configurations for a part of the configuration of each embodiment. Each of the above-described configurations, functions, processing units, processing means, and the like may be realized by hardware by designing a part or all of them with, for example, an integrated circuit. Each of the above-described configurations, functions, and the like may be realized by software by interpreting and executing a program that realizes each function by the processor. Information such as programs, tables, and files for realizing each function can be stored in a recording device such as a memory, a hard disk, or an SSD (Solid State Drive), or a recording medium such as an IC card, an SD card, or a DVD.

10 ... Control device,
20: Host server,
30 ... physical server,
40 ... Hypervisor,
45 ... Virtual control server schedule table,
50 ... Virtual control server,
60 ... Controlled device

Claims (10)

A program control device that periodically executes a plurality of programs,
An input unit for receiving an execution cycle of a newly executed program and an execution time of the executed program;
A program that is newly executed based on the input execution cycle and execution time so that the execution time zone of another program already scheduled for execution and the execution time zone of the program to be newly executed do not overlap in the CPU having the same identifier A schedule unit that generates an execution schedule including an identifier of a CPU that executes the program, an execution timing on a cycle of a program to be newly executed, and
An execution control unit that controls execution of the program based on the execution schedule ,
The program relates to a virtual control server, and in the virtual control server, it is determined whether or not the application that is controlled by the virtual control server is being processed. A program control apparatus that notifies that a virtual control server can be executed. The program control device according to claim 1,
The input unit inputs additional information of the newly executed program,
The program control apparatus according to claim 1, wherein the schedule unit determines an interval for executing the newly executed program and another program already scheduled for execution based on a predetermined criterion based on the additional information. The program control device according to claim 2,
The program is a program for controlling the operation of the railway, and the additional information given to the program is information about the railway line of the controlled object,
The program control apparatus according to claim 1, wherein the schedule unit schedules the program to be executed by different CPUs in line units. The program control device according to claim 1,
The scheduling unit changes the execution schedule so that the program scheduled to be executed by the first CPU is executed by the second executable CPU, and schedules the program to be continuously executed by the first CPU. The program control apparatus characterized by the above-mentioned. The program control apparatus according to claim 4, wherein
The input unit inputs additional information of the newly executed program,
The program control apparatus according to claim 1, wherein the schedule unit determines an interval for executing the newly executed program and another program already scheduled for execution based on a predetermined criterion based on the additional information. A program control method for periodically executing a plurality of programs,
The input unit receives the execution cycle of the program to be newly executed and the execution time of the program to be executed,
Based on the execution cycle and execution time input by the schedule unit, the execution time zone of other programs already scheduled for execution and the execution time zone of the newly executed program will not be duplicated in CPUs with the same identifier. Generate an execution schedule that includes the identifier of the CPU that executes the program to be executed and the execution timing on the cycle of the newly executed program,
The execution control unit controls the execution of the program based on the execution schedule ,
The program relates to a virtual control server, and in the virtual control server, it is determined whether or not the application that is controlled by the virtual control server is being processed. A program control method for notifying that a virtual control server can be executed. The program control method according to claim 6, wherein
The input unit inputs additional information of the newly executed program,
The program control method characterized in that the schedule unit determines an interval for executing the newly executed program and another program already scheduled for execution based on a predetermined criterion based on the additional information. The program control method according to claim 7,
The program is a program for controlling the operation of the railway, and the additional information given to the program is information about the railway line of the controlled object,
The program control method according to claim 1, wherein the schedule unit schedules to be executed by a different CPU for each line section. The program control method according to claim 6, wherein
The scheduling unit changes the execution schedule so that the program scheduled to be executed by the first CPU is executed by the second executable CPU, and schedules the program to be continuously executed by the first CPU. The program control method characterized by the above-mentioned. The program control method according to claim 9, wherein
The input unit inputs additional information of the newly executed program,
The program control method characterized in that the schedule unit determines an interval for executing the newly executed program and another program already scheduled for execution based on a predetermined criterion based on the additional information.

Images