JP2011243162A - Quantity control device, quantity control method and quantity control program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To increase the number of operating servers when needed to address an increase in load while suppressing resource usage of standby servers in a distributed processing system where the number of operating servers is varied in accordance with the resource usage of the servers.SOLUTION: The present invention calculates expected time until resource usage reaches a threshold which is an index to change a standby server to an operating server, selects a standby method to enable the standby server to be generated within the calculated expected time, and generates the standby server with the standby method.

Description

  The present invention relates to a distributed processing system that distributes processing to a plurality of servers, in which the number of servers that are operated is changed according to the resource usage of the server.

  In a distributed processing system, a load distribution apparatus distributes processing to a plurality of servers and distributes the load to each server. In particular, in some distributed processing systems, a load distribution apparatus responds to an increase or decrease in load by controlling the number of operating servers according to the overall load status.

  In addition, by using a virtualization technique, each server can be a virtual server, and server generation and deletion can be performed in a short time. Also, by making each server a virtual server, a new server can be generated without preparing new hardware. Therefore, in an environment in which a plurality of systems are mixed, control such as allocating more resources to a system with a temporarily high load is possible.

  In a distributed processing system, there is a method of preparing a standby server in advance and switching the standby server to an active server when it becomes necessary (see Patent Document 1) in order to cope with a sudden increase in load. However, since the standby server is a server that is not normally used, preparing the standby server uses useless resources when no increase in load occurs.

International Publication WO2006 / 040811

In Patent Document 1, a method for preparing a standby server is uniquely determined, and it is only possible to select whether the standby server of the method is activated or not activated. If the standby server is prepared in a state where it is activated in advance, it is possible to quickly cope with an increase in load, but if no increase in load occurs, useless resources are used. On the other hand, if the standby server is prepared without being activated, it is possible to suppress the use of resources when no increase in load occurs, but the response to the increase in load may be delayed.
An object of the present invention is to increase the number of active servers at a necessary timing while suppressing resource usage by a standby server, and to cope with an increase in load.

The number control device according to the present invention,
A distributed processing system that distributes processing to multiple servers, and the number of active servers in the distributed processing system in which the number of active servers is changed according to the amount of resource used, which is the amount of server resources used Is a unit control device that controls
A time storage unit that stores a generation time required to generate a standby server waiting for switching to the active server from a server in a predetermined setting state in a storage device for each standby method of the standby server;
A predicted time calculation unit that calculates a first predicted time until the resource usage reaches a first threshold that is set in advance by a processing device;
A standby method selection unit that selects a standby method that is shorter than the first predicted time calculated by the predicted time calculation unit by the generation time stored by the time storage unit by a processing device;
A standby server generation unit that generates a standby server of the standby method selected by the standby method selection unit from the server in the predetermined setting state by a processing device;
And a switching unit that operates the standby server generated by the standby server generation unit and switches to the active server when the resource usage amount reaches the first threshold.

  The number control device according to the present invention calculates a standby time of a standby server that can be generated within the calculated predicted time by calculating a predicted time until the resource usage reaches a threshold value that serves as an index for switching the standby server to an active server. Select and create a standby server for that standby method. Therefore, a standby server can be prepared at the time when switching to an active server is required, and the number of active servers can be increased at a necessary timing to cope with an increase in load.

1 is a configuration diagram of a server resource control system according to Embodiment 1. FIG. The figure which showed the structure of each virtual server 3 shown in FIG. 1 logically. The figure which shows the example of the resource provision time management table 26 which concerns on Embodiment 1. FIG. The figure which showed the example of the process (generation procedure) which a standby server produces | generates by the system shown in each standby system 26-1, and the process (switching procedure) at the time of changing from a standby server to an active server. FIG. 6 is a diagram showing an example of a resource management table 27 according to the first embodiment. 5 is a flowchart showing a flow of operations when the load of the resource management / control server 6 according to the first embodiment is increased. Explanatory drawing of the prediction calculation of the future load increase which concerns on Embodiment 1. FIG. The figure which shows the example of the resource management table 27 which concerns on Embodiment 2. FIG. 9 is a flowchart showing an operation flow when the load of the resource management / control server 6 according to the second embodiment increases. 9 is a flowchart showing an operation flow when the load of the resource management / control server 6 according to the second embodiment is reduced. FIG. 10 shows an example of a resource management table 27 according to the third embodiment. 10 is a flowchart showing a flow of operations at the start of processing for each server group of the resource management / control server 6 according to the third embodiment. FIG. 6 is a configuration diagram of a server resource control system according to a fourth embodiment. 10 is a flowchart showing a flow of operations when the active server of the resource management / control server 6 according to the fourth embodiment is stopped. Explanatory drawing of the prediction calculation of the future load rise which concerns on Embodiment 4. FIG. The figure which shows an example of the hardware constitutions of the physical computer 1, the resource management and control server 6, the load distribution apparatus 5, and the terminal 8. FIG.

Hereinafter, embodiments of the invention will be described with reference to the drawings.
In the following description, the processing device is a CPU 911 or the like which will be described later. The storage device is a storage device such as a ROM 913, a RAM 914, or a magnetic disk 920, which will be described later. That is, the processing device and the storage device are hardware.

Embodiment 1 FIG.
In this embodiment, a method for determining a standby method of a standby server prepared when the load increases will be described.

FIG. 1 is a configuration diagram of a server resource control system according to the first embodiment.
As shown in FIG. 1, the server resource control system includes a plurality of physical computers 1 (1-1, 1-2,...), A load distribution device 5, and a resource management / control server 6 (number control device). A plurality of terminals 8 and a storage 10. Each physical computer 1 is connected to the load balancer 5 and the resource management / control server 6 via a communication line 4. Each physical computer 1 and the storage 10 are connected via a communication line 9. The load balancer 5 and each terminal 8 are connected via the Internet / intranet 7.

Each physical computer 1 includes virtualization software 2 (2-1, 2-2,...). The virtualization software 2 is software that realizes server virtualization. In each physical computer 1, a server group constituting the system is constructed as virtual servers 3 (3-1 to 3-10,...) By the virtualization software 2.
In the physical computer 1-1, the Web 1-1 virtual server, the Web 2-3 virtual server, the App 1-1 virtual server, the DB-1 virtual server, and the Web 2-2 virtual server are constructed as the virtual server 3 by the virtualization software 2-1. Has been. In the physical computer 1-2, the Web 1-2 virtual server, the Web 2-1 virtual server, the App 2-1 virtual server, the Web 1-3 virtual server, and the App 1-2 virtual server are converted into the virtual server 3 by the virtualization software 2-2. Is built as.
Each virtual server 3 provides various services. Here, as the virtual server 3 that provides the function of Web1, there are a Web1-1 virtual server, a Web1-2 virtual server, and a Web1-3 virtual server. Further, as the virtual server 3 that provides the Web2 function, there are a Web2-1 virtual server, a Web2-2 virtual server, and a Web2-3 virtual server. Moreover, as the virtual server 3 that provides the function of App1, there are an App1-1 virtual server and an App1-2 virtual server. Further, as the virtual server 3 that provides the function of App2, there is an App2-1 virtual server. As a virtual server 3 having a DB function, there is a DB-1 virtual server.
Of these virtual servers, the Web 1-3 virtual server and the Web 2-3 virtual server are standby servers, and the others are active servers. Each active server is assigned available resources.
In the following description, the virtual server 3 that provides the same function is referred to as a server group. For example, the Web1-1 virtual server, the Web1-2 virtual server, and the Web1-3 virtual server are one server group, and the App1-1 virtual server and the App1-2 virtual server are one server group.

  The communication line 4 is, for example, a network such as an intranet or a LAN (Local Area Network). Although a single network is described here, a plurality of networks can be prepared using a management network, a business network, or the like.

The load distribution device 5 distributes requests for a certain function to a plurality of virtual servers 3 having the function, and transmits the request to distribute the processing of the same function, thereby realizing a function of improving the processing capability. It is a device.
As the load balancer 5, when an appliance server is introduced, when it is realized by software, a function of a DNS (Domain Name System) server is used, etc., any implementation method may be used. Absent.

The resource management / control server 6 is a server that controls the number of active servers in accordance with the resource usage of the virtual server 3. The resource management / control server 6 includes a resource information collection / accumulation unit 21, a load trend determination unit 22, a role change unit 23, a resource provision method determination unit 24, a standby generation / deletion unit 25, a resource provision time management table 26, resource management A table 27 is provided.
The resource information collection / accumulation unit 21 periodically collects resource usage information of each virtual server 3 and resource amount information allocated to each server group, and accumulates them in a storage device of the resource management / control server 6. . Further, the resource information collection / accumulation unit 21 determines whether or not the resource usage amount of the server group exceeds the first threshold defined for each server group by the processing device.
Based on the resource usage information collected by the resource information collection / accumulation unit 21, the load trend determination unit 22 performs a prediction calculation of a future increase in load by the processing device using regression analysis or the like.
The role changing unit 23 changes the standby server to the active server or changes the active server to the standby server by the processing device.
The resource provision method determination unit 24 generates a load increase tendency derived by the load trend determination unit 22 performing prediction calculation, information on the amount of resources allocated to a server group that needs to generate a standby server, and a standby server. The processing device determines an optimal standby server standby method based on the time required for the standby server to change to the active server and the resource amount information required for the standby server.
The standby generation / deletion unit 25 generates a standby server or deletes the standby server based on the standby method derived by the resource provision method determination unit 24 by the processing device.
The resource provision time management table 26 is a table (storage device) in which the generation time of each standby method, the time required to change a standby server to an active server, and resource usage information are stored.
The resource management table 27 is a table (storage device) that stores information such as configuration information and threshold values of each server group.

  The Internet / intranet 7 is a network used when making a request to each server group, and is the Internet, an intranet, a LAN, a dedicated line, or the like.

  The terminal 8 sends a request to each server group via the Internet / intranet 7.

  The communication line 9 is a network SAN (Storage Area Network) through which each physical computer accesses the storage.

  The storage 10 is a storage that stores a boot disk image of the virtual server 3 running on each physical computer 1 and a template that is a basis for generating a standby server. In particular, the virtual server disk image 11 is a boot disk image of each virtual server 3. The virtual server template 12 is a template that is a basis for generating a standby server for each server group.

FIG. 2 is a diagram logically showing the configuration of each virtual server 3 shown in FIG. As shown in FIG. 2, the system realized by each virtual server 3 is an environment in which two systems having Web1 and Web2 as front ends are mixed.
A request from the terminal 8 is delivered to the load balancer 5 via the Internet / intranet 7. If the request is for Web1, the request is sent from the load balancer 5 to either Web1-1 or Web1-2. Further, a request is sent to App1-1 or App1-2 via the load balancer 5 as necessary. Requests are sent to DB-1 as necessary. Then, the processing result returns in the reverse order of the order in which the requests are sent.
Similarly, if the request is for Web 2, the request is sent from the load balancer 5 to either Web 2-1 or Web 2-2. Further, a request is sent to App 2-1 through the load balancer 5 as necessary. Then, the processing result returns in the reverse order of the order in which the requests are sent.

FIG. 3 is a diagram illustrating an example of the resource provision time management table 26 according to the first embodiment.
The resource provision time management table 26 includes a standby method 26-1, a switching time 26-2 (time required from standby to operation), a generation time 26-3 (time for generation), and a resource usage rank 26-4. It is a table having items.
The standby method 26-1 stores a method name for identifying a method for preparing a standby server. In this example, suspending the standby server in a suspended state, a snapshot that saves the standby server state, a boot with a standby server boot image, and a template that remains in the state of the template on which the standby server is based An example prepared as is shown.
The switching time 26-2 stores the time required to change the standby server prepared in each standby method to an active server.
The generation time 26-3 stores the time required to generate the standby server of each standby method.
The resource usage ranking 26-4 stores the ranking of the amount of resources (resource usage) used by the standby server when a standby server is prepared in each standby method. In this example, the larger the resource usage, the smaller the number.

FIG. 4 is a diagram showing an example of processing (generation procedure) for generating a standby server by the method shown in each standby method 26-1 and processing (switching procedure) when changing from a standby server to an active server. is there.
Here, it is assumed that a template in which installation of an OS, an application, and the like for generating a standby server of each server group is completed is prepared in advance.

(suspend)
The generation procedure is the following (1) to (3). (1) Copy a boot image from a template and generate a boot disk. (2) Start with the boot image of the boot disk. (3) Set to a temporary stop state.
The switching procedure is the following (1) and (2). (1) Release the pause state and operate. (2) After entering the operating state, register in the load balancer 5.

(snap shot)
The generation procedure is the following (1) to (4). (1) Copy a boot image from a template and generate a boot disk. (2) Start with the boot image of the boot disk. (3) Generate snapshot information. (4) After the generation of the snapshot information is completed, the generated standby server is shut down.
The switching procedure is the following (1) and (2). (1) Start from a snapshot. (2) After entering the operating state, register in the load balancer 5.

(boot)
The generation procedure is the following (1). (1) Copy a boot image from a template and generate a boot disk.
The switching procedure is the following (1) and (2). (1) Start with the boot image of the boot disk. (2) After entering the operating state, register in the load balancer 5.

(template)
Since a template is assumed to be prepared, there is no generation procedure.
The switching procedure is from the following (1) to (3). (1) Copy a boot image from a template and generate a boot disk. (2) Start with the boot image of the boot disk. (3) After entering the operating state, register in the load balancer 5.

  As described above, since the generation procedure and switching procedure differ for each standby method, the time required for generating the standby server for each standby method, the time required for switching from the standby server to the active server, and the resources used by the standby server The amount used is different. Therefore, these pieces of information are stored in the resource provision time management table 26, and an optimum standby method is derived.

  In the switching procedure, a process of always registering in the load balancer 5 is performed at the end. In order to shorten the time required for the process of switching the standby server to the active server, an IP address that can be assigned to each server group is secured in advance, and the IP address secured when the standby server is generated is assigned to the standby server. It may be assigned. In this case, when the standby server is switched to the active server, the IP address assigned to the standby server may be registered in the load balancer 5. That is, when switching from the standby server to the active server, the load balancer 5 identifies which server group's active server is added with which IP address simply by registering the IP address in the load balancer 5. Is possible.

FIG. 5 is a diagram illustrating an example of the resource management table 27 according to the first embodiment.
The resource management table 27 includes a server group name 27-1, an active server 27-2, a standby server 27-3, a standby method 27-4, a template name 27-5, a maximum number 27-6, a first threshold 27-7, It is a table which has the item of the 2nd threshold value 27-8 and the 3rd threshold value 27-9.
The server group name 27-1 stores a name for identifying each server group.
The active server 27-2 stores a name for identifying a server operating as an active server in each server group.
The standby server 27-3 stores a name for identifying a standby server prepared at the present time.
In the standby method 27-4, the name of the standby method of the prepared standby server is stored. The standby method 27-4 stores one of the method names recorded in the standby method 26-1 of the resource provision time management table 26.
The template name 27-5 stores a name for identifying a template that is a basis for generating the standby server.
The maximum number 27-6 stores the maximum number of servers that can operate in each server group.
As the first threshold 27-7, the second threshold 27-8, and the third threshold 27-9, resource thresholds are stored. The first threshold value 27-7 stores a value used as a trigger for starting the change from the standby server to the active server. The second threshold value 27-8 is a value used as the upper limit value of the resource usage amount of each server group, and stores a value that should prevent a state exceeding this value from continuing for a long time. The third threshold 27-9 stores a value used as a trigger for reducing the number of active servers by reducing the resource usage. In each threshold value, the ratio of the resource usage amount to the resource amount allocated to the server group is stored as a threshold value.

FIG. 6 is a flowchart showing an operation flow when the load of the resource management / control server 6 according to the first embodiment increases. The resource management / control server 6 executes the process shown in FIG. 6 for each server group.
In (ST101), the resource information collection / accumulation unit 21 periodically collects resource usage information indicating the resource usage of each virtual server 3, and stores the first threshold value 27-7 stored in the resource management table 27. It is determined by the processing device whether or not the threshold is exceeded. Here, the resource usage information collected by the resource information collection / accumulation unit 21 is accumulated in the storage device of the resource management / control server 6.
If it is determined that the first threshold value 27-7 has been exceeded, the process proceeds to (ST102). On the other hand, if it is determined that the first threshold value 27-7 is not exceeded, (ST101) is executed again after a predetermined time has elapsed.

In (ST102), the role changing unit 23 determines from the information stored in the standby server 27-3 of the resource management table 27 whether the standby server is prepared by the processing device.
If a standby server is prepared, the process proceeds to (ST105). On the other hand, if no standby server is prepared, the process proceeds to (ST103).

In (ST103), whether or not the role changing unit 23 has reached the maximum number of operating servers by the processing device from the information stored in the operating servers 27-2 and the maximum number 27-6 of the resource management table 27. Determine whether.
If it is determined that the maximum number has been reached, the operation server cannot be added, and the process returns to (ST101). If it is determined that the maximum number has not been reached, the process proceeds to (ST104).

In (ST104), when the standby server is not prepared even though the maximum number has not been reached, since the standby server is being generated, the role changing unit 23 has completed the generation of the standby server by the processing device. It is determined whether or not.
If it is determined that the generation has not been completed, (ST104) is executed again after a predetermined time has elapsed. On the other hand, if it is determined that the generation has been completed, the process proceeds to (ST105).

In (ST105), the role changing unit 23 switches the standby server to the active server based on the information stored in the standby method 27-4 of the resource management table 27. Further, the role changing unit 23 updates the information stored in the active server 27-2, standby server 27-3, and standby method 27-4 in the resource management table 27.
Then, the process proceeds to (ST106).

In (ST106), the load tendency determination unit 22 performs a prediction calculation of a future load increase by the processing device.
FIG. 7 is an explanatory diagram of a prediction calculation of a future load increase according to the first embodiment. In FIG. 7, each plot is a resource usage amount indicated by the resource usage information collected in (ST101) and accumulated in the storage device at a past predetermined time (time t0). From this resource usage amount (distribution of resource usage status), a method such as regression analysis is used to predict the future load increase. For example, an approximate expression such as y = ax + b (y: resource use amount, x: time, a, b: values calculated from the resource use amount) is derived, and prediction calculation is performed from the derived approximate expression.
Then, the process proceeds to (ST107).

In (ST107), the resource providing method determination unit 24 uses the processing device to determine the resource usage based on the load increase trend derived by the load trend determination unit 22 and the resource amount currently assigned to the target server group. The time until the amount reaches the first threshold value 27-7 (t1 in FIG. 7, the first predicted time) is calculated. Also, the resource provision method determination unit 24 uses the processing device to wait for the resource usage amount to reach the second threshold value 27-8 after reaching the first threshold value 27-7 (t2, second predicted time in FIG. 7). Calculate
Note that the resource amount currently assigned to the target server group is added in (ST105) to the total amount of resources assigned to each active server until the active server is added in (ST105). This is the amount of resources added to the amount of resources allocated to the active server.

In (ST108), the resource provision method determination unit 24 compares the time stored in the generation time 26-3 of the resource provision time management table 26 with the first predicted time calculated in (ST107) by the processing device. . Further, the resource provision method determination unit 24 compares the time stored in the switching time 26-2 of the resource provision time management table 26 with the second predicted time calculated in (ST107) by the processing device. Whether the resource management table 27 has a standby method in which the generation time 26-3 is shorter than the first prediction time and the switching time 26-2 is shorter than the second prediction time. judge.
If it is determined that there is a standby method that satisfies both times, the process proceeds to (ST109). On the other hand, if it is determined that there is no standby method that satisfies both times, the process proceeds to (ST110).

In (ST109), the resource provision method determination unit 24 determines whether there is only one method or a plurality of standby methods satisfying both times by the processing device.
If it is determined that there is only one method, the process proceeds to (ST112). On the other hand, if it is determined that there are a plurality, the process proceeds to (ST111).

In (ST110), the resource provision method determination unit 24 determines that the sum of the time stored in the switching time 26-2 and the time stored in the generation time 26-3 of the resource provision time management table 26 is the largest by the processing device. Choose a smaller standby method.
Then, the process proceeds to (ST112).

  In (ST111), the resource provision method determination unit 24 selects a standby method with a low value (the resource use amount is small) stored in the resource use amount ranking 26-4 of the resource provision time management table 26 by the processing device. To do.

In (ST112), the standby generation / deletion unit 25 generates a standby server of the standby method determined by the resource provision method determination unit 24 by the processing device. Then, the standby generation / deletion unit 25 updates the information of the standby server 27-3 and the standby method 27-4 in the resource management table 27.
Then, the process returns to (ST101).

As described above, the resource management / control server 6 selects a standby method capable of adding an active server (adding resources) within a required time based on the load tendency. In addition, when there are a plurality of standby methods in which an active server can be added within the required time, a standby method that uses fewer resources is selected.
Therefore, it is possible to cope with an increase in load by changing the standby server to an active server at a necessary timing while suppressing the resource usage of the standby server.
In particular, even if the same server group has the same load trend, if the number of active servers is small, the total amount of resources allocated to the server group is small, so it must be added immediately and the number of active servers is large. However, since the total amount of resources allocated to the server group is large, immediate addition is not necessary. According to the resource management / control server 6, an active server can be added in accordance with such a tendency.

In the above description, the configuration of the virtual server is described. However, the configuration can be applied to the configuration of the physical server.
In the above description, an example of a linear expression based on regression analysis is shown as a method for determining the load tendency. However, another method may be used as long as the method can predict the load tendency.
In the above description, suspend, snapshot, boot, and template are shown as examples of the standby method. However, the present invention is not limited to these methods.

In other words, the server resource control method according to the first embodiment distributes processing to a plurality of grouped servers, and in a system that realizes an increase in processing amount by adding servers,
Assigned to the time information for generating the standby server, the time information for changing the role from the standby server to the active server, the load fluctuation trend derived from the resource usage history information of the active server, and the active server group at that time Based on the time estimate until reaching the threshold that changes depending on the amount of resources that are available, the standby server can be used within the time required to change the resource amount of the active server and the standby server is selected with the least amount of resource usage. Is generated.

Embodiment 2. FIG.
In this embodiment, a method of recording a standby method of a standby server selected at the time of load increase and determining a standby method of a standby server at the time of load reduction based on the recorded standby method will be described.

FIG. 8 is a diagram illustrating an example of the resource management table 27 according to the second embodiment.
In the resource management table 27 shown in FIG. 8, an initial number 27-10 and a selection standby method 27-11 are added to the resource management table 27 shown in FIG. The rest is the same as the resource management table 27 shown in FIG.
The initial number 27-10 stores the initial number of active servers. Here, it is assumed that the number of operating servers cannot be reduced to a smaller number than the initial number of operating servers.
The selection standby method 27-11 stores which method has been selected as the standby method for each number of active servers. For example, in FIG. 8, in the case of the Web1 server group, the standby server standby method in the case of two active servers is suspended, and the standby server standby method in the case of three is boot, and four Indicates that the standby method of the standby server in this case is boot.

FIG. 9 is a flowchart showing an operation flow when the load of the resource management / control server 6 according to the second embodiment increases.
The operations from (ST101) to (ST112) in FIG. 9 are the same as the operations shown in FIG. 6 described in the first embodiment.
In (ST201), the resource providing method determination unit 24 stores the selected standby method in an item according to the number of active servers in the selected standby method 27-11 of the resource management table 27.

FIG. 10 is a flowchart showing an operation flow when the load of the resource management / control server 6 according to the second embodiment is reduced. The resource management / control server 6 executes the process shown in FIG. 10 for each server group.
In (ST202), whether or not the resource information collection / accumulation unit 21 periodically collects the resource usage information of each virtual server 3 and falls below the third threshold value 27-9 stored in the resource management table 27. Is determined by the processing device. Here, the resource usage information collected by the resource information collection / accumulation unit 21 is accumulated in the storage device of the resource management / control server 6.
If it is determined that the value falls below the third threshold 27-9, the process proceeds to (ST203). On the other hand, if it is determined that it is not lower than the third threshold 27-9, (ST202) is executed again after a predetermined time has elapsed.

In (ST203), the resource provision method determination unit 24 determines whether the number of active servers is the initial number by the processing device from the information stored in the active servers 27-2 and the initial number 27-10 of the resource management table 27. Determine whether.
If it is determined that the number is not the initial number, the process proceeds to (ST204). On the other hand, if it is determined that the number is the initial number, the number of active servers cannot be reduced, so the process returns to (ST202).

In (ST204), the resource provision method determination unit 24 calculates the current number of active servers from the information stored in the active server 27-2 of the resource management table 27 by the processing device. Then, the resource provision method determination unit 24 reads the standby method stored in the item of the number reduced by one from the number of active servers calculated in the selection standby method 27-11 of the resource management table 27.
Then, the process proceeds to (ST205).

  In (ST205), the role changing unit 23 switches one of the currently active servers to the standby server of the standby method read in (ST204) by the processing device. Then, the role changing unit 23 updates information on the active server 27-2, the standby server 27-3, and the standby method 27-4 in the resource management table 27.

  In (ST206), the standby generation / deletion unit 25 deletes the standby server prepared before switching to the standby server in (ST205) by the processing device.

  As described above, when the load is increasing, the standby method can be selected depending on the time and the resource usage state. However, when the load is decreasing, it is difficult to select the standby method because the next upward trend cannot be predicted. Therefore, by recording the standby method selected when the load increases, and selecting the standby method recorded when the load decreases, the optimal standby method for the system can be selected.

  In the flow shown in FIG. 9, the standby method in the case of the same number of active servers is always updated to the latest information. However, a standby method that uses the least resources among the past information, or a standby method that switches the standby server to the active server at the highest speed may be recorded.

  In other words, the server resource control method according to the second embodiment records the standby server standby method prepared after adding the active server in association with the number of active servers as the load increases, and the active server when the load decreases By preparing a standby method corresponding to the number of active servers in the recorded method when the number of servers is reduced, the standby server standby method at the time of load reduction can be used as it is and the method selected at the time of load increase can be used as it is. This makes it possible to select a standby method suitable for the system without performing a complicated calculation.

Embodiment 3 FIG.
In this embodiment, a method for changing the operation of preparing a standby server according to the role of the server group will be described.

FIG. 11 is a diagram illustrating an example of the resource management table 27 according to the third embodiment.
In the resource management table 27 shown in FIG. 11, a front group 27-12 is added to the resource management table 27 shown in FIG. The rest is the same as the resource management table 27 shown in FIG.
When there is another server group that operates as a front end of the server group indicated by the name of the server group name 27-1, the front group 27-12 stores the name of the server group. If there is no other server group operating as a front end, information of “none” is stored.
The server group operating on the front end is a server group operating on the terminal 8. For example, as shown in FIG. 2, since the server group of Web1 operates from the terminal 8 rather than the server group of App1, the server group of Web1 is a server group that operates as a front end of the server group of App1. In this case, the App1 server group is referred to as a server group that operates as a back end of the Web1 server group.

FIG. 12 is a flowchart showing an operation flow at the start of processing for each server group of the resource management / control server 6. As a premise, a standby server is not prepared in the back-end server group, and a standby method to be prepared in the standby method 27-4 of the resource management table 27 is set.
In (ST301), based on the information stored in the front group 27-12 of the resource management table 27, the resource provision method determining unit 24 determines whether the server group includes a server group that operates on the front end based on the information. Judging.
If it is determined that there is no server group operating on the front end, the process proceeds to (ST101) shown in FIGS. 6 and 9, and thereafter, the processes shown in FIGS. 6 and 9 are executed. On the other hand, if it is determined that the front-end server group exists, the process proceeds to (ST302).

In (ST302), the resource provision method determination unit 24 determines whether or not there is a standby server in the front-end server group by the processing device from the information stored in the standby server 27-3 of the resource management table 27. .
If it is determined that there is a standby server, the process proceeds to (ST101) shown in FIGS. 6 and 9, and thereafter, the processes shown in FIGS. 6 and 9 are executed. On the other hand, if it is determined that there is no standby server, the process proceeds to (ST303).

In (ST303), the resource provision method determination unit 24 determines whether a standby server has been added to the front-end server group by the processing device. Note that the addition of the standby server means that a standby server has been generated separately from the standby server prepared in the initial state. For example, if the number of active servers is larger than the initial number 27-10 of the resource management table 27 and a standby server is prepared, it can be determined that a standby server has been added.
If it is determined that it has not been added, (ST303) is executed again after a predetermined time has elapsed. On the other hand, if it is determined that it has been added, the process proceeds to (ST304).

  In (ST304), the standby generation / deletion unit 25 generates the standby server of the standby method recorded in the standby method 27-4 of the resource management table 27 by the processing device.

  In (ST305), the standby generation / deletion unit 25 records the generated standby method in an item corresponding to the number of active servers in the selection standby method 27-11 of the resource management table 27.

If resources are not added to the front-end server group, even if resources are added to the back-end server group, the amount of processing cannot be increased due to the front-end server group becoming a bottleneck.
Therefore, as described above, the server group serving as the back end has a standby server after it is determined that the load on the front end server group has increased and resources have been added from the initial state (an active server has been added). Prepare for. Thereby, it is possible to prevent an unnecessary increase in the amount of resource used by the standby server.

  In other words, the server resource control method according to the third embodiment is based on the hierarchical structure information between groups, and when the load increases, the number of active servers in the front group is added and By starting generation of a standby server of a certain group, a standby server on the back-end side is not prepared in advance, thereby suppressing resource reservation for the standby server.

Embodiment 4 FIG.
In this embodiment, an operation when an active server is stopped due to a failure or the like will be described.

FIG. 13 is a configuration diagram of a server resource control system according to the fourth embodiment. A failure monitoring unit 31 is added to the configuration diagram shown in FIG. The configuration other than the added failure monitoring unit 31 is the same as that in FIG.
The failure monitoring unit 31 monitors the operating state of the active server in each server group by the processing device.

FIG. 14 is a flowchart showing an operation flow when the active server of the resource management / control server 6 according to the fourth embodiment is stopped.
In (ST401), the failure monitoring unit 31 monitors the operating state of the active server in the server group by the processing device, and determines whether the active server has stopped due to a failure or the like.
If it is determined that it has not stopped, (ST401) is executed again after a predetermined time has elapsed. On the other hand, if it is determined that the process has been stopped, the process proceeds to (ST402).

In (ST402), based on the information stored in the standby server 27-3 of the resource management table 27 for the server group in which the active server is determined to have been stopped in (ST401), the failure monitoring unit 31 uses the processing device to It is determined whether or not a standby server is prepared.
If it is determined that it is prepared, the process proceeds to (ST408). On the other hand, if it is determined that it is not prepared, the process proceeds to (ST403).

In (ST403), whether or not the role changing unit 23 has reached the maximum number of active servers by the processing device from the information stored in the active servers 27-2 and the maximum number 27-6 of the resource management table 27. Determine whether.
If it is determined that the maximum number has been reached, the process proceeds to (ST404). On the other hand, if it is determined that the maximum number has not been reached, the process proceeds to (ST407). Here, since the active server has stopped, it is possible to add a new active server even if the maximum number is reached.

In (ST404), the resource provision method determination unit 24 selects the standby method with the smallest sum of the switching time 26-2 and the generation time 26-3 of the resource provision time management table 26 by the processing device. Then, the process proceeds to (ST405).
In (ST405), the standby generation / deletion unit 25 generates a standby server of the standby method determined by the resource provision method determination unit 24 by the processing device.
In (ST406), the role changing unit 23 changes the generated standby server to an active server by the processing device. Thereby, the number of active servers returns to the state (maximum number) before the active servers are stopped.

In (ST407), since the standby server is being generated when the maximum number has not been reached, the role changing unit 23 determines whether or not the standby server has been generated by the processing device.
If it is determined that the generation has not been completed, (ST407) is executed again after a predetermined time has elapsed. On the other hand, if it is determined that the generation has been completed, the process proceeds to (ST408).

In (ST408), the role changing unit 23 switches the standby server to the active server based on the information stored in the standby method 27-4 of the resource management table 27. Thereby, the number of active servers returns to the state before the active servers are stopped. Further, the role changing unit 23 updates the information stored in the active server 27-2, standby server 27-3, and standby method 27-4 in the resource management table 27.
Then, the process proceeds to (ST409).

In (ST409), the load tendency determination unit 22 performs prediction calculation of a future load increase by the processing device.
FIG. 15 is an explanatory diagram of a prediction calculation of a future load increase according to the fourth embodiment. In FIG. 15, there are two cases: a case where a standby server is generated when a failure occurs (when the answer is YES in ST402) and a case where a standby server is being generated (via ST407). Show.
When the number of active servers returns to the original configuration (in FIG. 15, when the generation of a standby server is completed when a failure occurs, when the failure occurs, and when the generation of a standby server is not completed when a failure occurs) , When the generation of the standby server is completed), the load trend determination unit 22 uses a method such as regression analysis from the distribution of resource usage in the past predetermined time (time t0) to predict the future load increase. I do. For example, an approximate expression such as y = ax + b (y: resource use amount, x: time, a, b: values calculated from the resource use amount) is derived, and prediction calculation is performed from the derived approximate expression.

  The processing from (ST410) to (ST415) is the same as the processing from (ST107) to (ST112) in FIG.

As described above, when a failure occurs in the active server and the amount of resources currently allocated decreases, the standby server is immediately changed to the active server, and a standby server is prepared for future resource increase. To do. As a result, it is possible to minimize the decrease time of the resource amount when a failure occurs, and to cope with an increase in load that may occur thereafter.
Also, the standby method generated when a failure occurs is not recorded. As a result, the standby method when the load is reduced can be provided with the normal standby method without being affected by the failure.

  In other words, the server resource control method according to the fourth embodiment immediately changes the standby server to the active server when the server is disconnected due to a failure or the like, and the resource usage history information of the active server at the time of the failure. Recalculate the load fluctuation trend derived from, and based on the time estimate for reaching the threshold that changes according to the amount of resources allocated to the active server group at that time, it can respond within the time required to change the resource amount of the active server In addition, the standby server is generated by selecting the standby server having the least amount of resource usage.

Next, the hardware configuration of the physical computer 1, the load balancer 5, the resource management / control server 6, and the terminal 8 in the embodiment will be described.
FIG. 16 is a diagram illustrating an example of a hardware configuration of the physical computer 1, the load balancer 5, the resource management / control server 6, and the terminal 8.
As shown in FIG. 16, the physical computer 1, the load distribution device 5, the resource management / control server 6, and the terminal 8 are a CPU 911 (Central Processing Unit, central processing unit, processing unit, arithmetic unit, micro unit that executes a program. A processor, a microcomputer, and a processor). The CPU 911 is connected to the ROM 913, the RAM 914, the LCD 901 (Liquid Crystal Display), the keyboard 902 (K / B), the communication board 915, and the magnetic disk device 920 via the bus 912, and controls these hardware devices. Instead of the magnetic disk device 920 (fixed disk device), a storage device such as an optical disk device or a memory card read / write device may be used. The magnetic disk device 920 is connected via a predetermined fixed disk interface.

  The ROM 913 and the magnetic disk device 920 are examples of a nonvolatile memory. The RAM 914 is an example of a volatile memory. The ROM 913, the RAM 914, and the magnetic disk device 920 are examples of a storage device (memory). The keyboard 902 and the communication board 915 are examples of input devices. The communication board 915 is an example of a communication device (network interface). Furthermore, the LCD 901 is an example of a display device.

  An operating system 921 (OS), a window system 922, a program group 923, and a file group 924 are stored in the magnetic disk device 920 or the ROM 913. The programs in the program group 923 are executed by the CPU 911, the operating system 921, and the window system 922.

The program group 923 includes “virtualized software 2”, “resource information collection / accumulation unit 21”, “load tendency determination unit 22”, “role change unit 23”, “resource provision method determination unit 24” in the above description. , Software, programs and other programs for executing the functions described as “standby generation / deletion unit 25”, “failure monitoring unit 31”, and the like are stored. The program is read and executed by the CPU 911.
In the file group 924, information, data, signal values, variable values, and parameters stored in the “resource provision time management table 26” and “resource management table 27” in the above description are stored in the “file” and “database”. Stored as each item. The “file” and “database” are stored in a recording medium such as a disk or a memory. Information, data, signal values, variable values, and parameters stored in a storage medium such as a disk or memory are read out to the main memory or cache memory by the CPU 911 via a read / write circuit, and extracted, searched, referenced, compared, and calculated. Used for the operation of the CPU 911 such as calculation / processing / output / printing / display. Information, data, signal values, variable values, and parameters are temporarily stored in the main memory, cache memory, and buffer memory during the operation of the CPU 911 for extraction, search, reference, comparison, calculation, calculation, processing, output, printing, and display. Is remembered.

In the above description, the arrows in the flowchart mainly indicate input / output of data and signals, and the data and signal values are recorded in a memory of the RAM 914, other recording media such as an optical disk, and an IC chip. Data and signals are transmitted online by a bus 912, signal lines, cables, other transmission media, and radio waves.
In addition, what is described as “to part” in the above description may be “to circuit”, “to device”, “to device”, “to means”, and “to function”. It may be “step”, “˜procedure”, “˜processing”. In addition, what is described as “˜device” may be “˜circuit”, “˜device”, “˜means”, “˜function”, and “˜step”, “˜procedure”, “ ~ Process ". Furthermore, what is described as “to process” may be “to step”. That is, what is described as “˜unit” may be realized by firmware stored in the ROM 913. Alternatively, it may be implemented only by software, or only by hardware such as elements, devices, substrates, and wirings, by a combination of software and hardware, or by a combination of firmware. Firmware and software are stored in a recording medium such as ROM 913 as a program. The program is read by the CPU 911 and executed by the CPU 911. That is, the program causes a computer or the like to function as the “˜unit” described above. Alternatively, the computer or the like is caused to execute the procedures and methods of “to part” described above.

  1 physical computer, 2 virtualization software, 3 virtual server, 4,9 communication line, 5 load balancer, 6 resource management / control server, 7 Internet / Intranet, 8 terminal, 10 storage, 11 virtual server disk image, 12 virtual Server template, 21 Resource information collection / accumulation unit, 22 Load trend determination unit, 23 Role change unit, 24 Resource provision method determination unit, 25 Standby generation / deletion unit, 26 Resource provision time management table, 27 Resource management table, 31 Failure Monitoring department.

Claims (9)

A distributed processing system that distributes processing to multiple servers, and the number of active servers in the distributed processing system in which the number of active servers is changed according to the amount of resource used, which is the amount of server resources used Is a unit control device that controls
A time storage unit that stores a generation time required to generate a standby server waiting for switching to the active server from a server in a predetermined setting state in a storage device for each standby method of the standby server;
A predicted time calculation unit that calculates a first predicted time until the resource usage reaches a first threshold that is set in advance by a processing device;
A standby method selection unit that selects a standby method that is shorter than the first predicted time calculated by the predicted time calculation unit by the generation time stored by the time storage unit by a processing device;
A standby server generation unit that generates a standby server of the standby method selected by the standby method selection unit from the server in the predetermined setting state by a processing device;
A number control device comprising: a switching unit that operates the standby server generated by the standby server generation unit and switches to the active server when the resource usage amount reaches the first threshold. The time storage unit further stores, for each standby method of the standby server, a switching time required to operate the standby server and switch to the active server,
The predicted time calculation unit further calculates a second predicted time until the resource usage reaches a second threshold in which a resource usage larger than the first threshold is set,
The standby mode selection unit selects the standby mode in which the generation time is shorter than the first prediction time and the switching time is shorter than the second prediction time. Number control device. The standby method selection unit, when there is no standby method in which the generation time is shorter than the first prediction time and the switching time is shorter than the second prediction time, the generation time and the switching time, The number control device according to claim 2, wherein the standby method with the shortest sum is selected. The standby method selection unit, when there are a plurality of standby methods that satisfy the condition that the generation time is shorter than the first prediction time and the switching time is shorter than the second prediction time, The number control device according to claim 2 or 3, wherein a standby method that uses a small amount of resources used by generating a standby server of the standby method is selected from among the standby methods that satisfy the condition. The number control device further includes:
A standby method storage unit for storing the standby method selected by the standby method selection unit;
The switching unit further waits for at least one active server when the resource usage is less than a third threshold for which the resource usage is less than the first threshold. The number control device according to claim 1, wherein the number control device is switched to a standby server of a standby method stored in a method storage unit. The plurality of servers are divided into a plurality of groups according to roles,
The number control device controls the number of active servers belonging to each group of the plurality of groups,
Whether the standby server generation unit generates a standby server belonging to the predetermined group according to whether there exists a standby server belonging to a front end group closer to the client than the predetermined group among the plurality of groups. 6. The number control device according to claim 1, wherein it is determined whether or not. When the active server is stopped, the switching unit operates the standby server to switch to the active server,
The predicted time calculation unit recalculates a first predicted time until the resource usage reaches the first threshold after the switching unit operates the standby server and switches to the active server,
The standby method selection unit selects the standby method in which the generation time stored in the time storage unit is shorter than the first prediction time recalculated by the prediction time calculation unit,
The number of standby servers according to any one of claims 1 to 6, wherein the standby server generation unit generates a standby server of the standby method selected by the standby method selection unit from the servers in the predetermined setting state. Control device. A distributed processing system that distributes processing to multiple servers, and the number of active servers in the distributed processing system in which the number of active servers is changed according to the amount of resource used, which is the amount of server resources used Is a unit control method for controlling
A prediction time calculation step in which the processing device calculates a first prediction time until the resource usage reaches a first threshold that is set in advance with a predetermined resource usage;
A standby time shorter than the first predicted time calculated in the predicted time calculation step is a time required for the processing device to generate a standby server waiting for switching to the active server from a server in a predetermined setting state. A standby method selection step for selecting a method;
A standby server generation step of generating a standby server of the standby method selected in the standby method selection step from the server in the predetermined setting state, the processing device;
A number control method comprising: a switching step of operating the standby server generated in the standby server generation step and switching to the active server when the resource usage amount reaches the first threshold value. A distributed processing system that distributes processing to multiple servers, and the number of active servers in the distributed processing system in which the number of active servers is changed according to the amount of resource used, which is the amount of server resources used Is a unit control program for controlling
A predicted time calculation process for calculating a first predicted time until the resource usage reaches a first threshold that is set in advance with a predetermined resource usage;
Select a standby method in which the generation time required to generate a standby server waiting for switching to the active server from a server in a predetermined setting state is shorter than the first predicted time calculated in the predicted time calculation process Standby method selection processing,
A standby server generation process for generating a standby server of the standby system selected in the standby system selection process from the server in the predetermined setting state;
A number control program for causing a computer to execute a switching process of operating a standby server generated in the standby server generation process and switching to an active server when the resource usage amount reaches the first threshold.

Images