KR102316749B1 - Virtual machine workload prediction method, virtual machine placement method and virtual machine placement device implementing the same - Google Patents

Abstract

A virtual machine placement method according to an embodiment of the present invention includes a physical server in which a virtual machine is installed using a virtual machine placement device, wherein the physical server includes a first physical server and a second physical server. In the virtual machine arrangement method, which is a method of disposing a plurality of the virtual machines on the first physical server and the second physical server in order to efficiently operate the calculating virtual machine workload information, which is information related to a virtual machine workload, which is a load of a physical server imposed by the virtual machine, by a predetermined workload calculation method based on operation-related information; calculating initial predicted virtual load information, which is information related to an initial predicted virtual load associated with a load of a physical server imposed by the virtual machine, predicted for a first predetermined period based on the virtual machine workload; calculating arrangement schedule information, which is information related to arrangement schedules of the virtual machines on the first physical server and the second physical server, by a predetermined arrangement schedule calculating method; displaying the arrangement schedule calculated by the predetermined arrangement schedule calculation method; and the virtual machines are arranged between the first physical server and the second so that the virtual machines are placed on the first physical server and the second physical server according to the arrangement schedule calculated by the predetermined arrangement schedule calculation method. and disposing while moving on a physical server, wherein the method for calculating the predetermined arrangement schedule includes calculating the arrangement schedule of the virtual machine when an objective function value is minimized based on the initial predicted virtual load information. , wherein the arrangement schedule is calculated so that the movement of the virtual machine between the physical servers is permitted during the first predetermined period, wherein the objective function is, during the first predetermined period, the initial predicted virtual load. A first objective function, which is a function related to the amount of change in the load of the physical server calculated by

Description

Virtual machine workload prediction method, virtual machine placement method, and virtual machine placement device implementing the same

The present invention relates to a virtual machine workload prediction method, a virtual machine placement method, and a virtual machine placement apparatus implementing the same, in order to efficiently operate a first physical server and a second physical server on which a plurality of virtual machines are respectively placed, and disposing a plurality of the virtual machines on the first physical server and the second physical server.

Data centers, which account for about 2% of the world's electricity consumption, are high-density and energy-consuming buildings that use the most electricity among single buildings, but they are inefficiently operating by adhering to conservative operations that prioritize stability. In the future, due to the spread of the Internet of Things, Big Data, and cloud technologies, it is expected that small data centers will be integrated into large-scale 'hyper-scale' and 'mega' data centers. Data center operation and management according to infrastructure expansion complexity is also expected to increase. Unnecessary servers are switched to idle or power saving mode, and available servers are operated as stably as possible to maximize data center operation efficiency and reduce power consumption to reduce operating costs. Proper placement of them is very important.

In a virtualization environment in which physical resources (Physical Machine, PM, physical server) are composed of logical resources (Virtual Machine, VM, virtual machine), a running virtual machine can be moved to another physical server, which is called live migration. This is an essential management method for efficient data center management. However, there are also disadvantages such as risk of data loss and a large portion of resources occupied by migration.

Here, existing studies (Japanese Patent Publication No. 5827594, 2015. 10. 23, registered) have migrated when a load greater than a threshold value is applied to any physical server based on a specific point in time, and the load of the loaded physical server is performed. A way to reduce the number of physical servers is to balance the physical servers. However, this method has a problem in that the server to which a specific virtual machine is migrated is highly likely to be overloaded, and as a result, a large number of migrations are executed as the migration process is executed again.

The problem to be solved by the present invention is to provide a virtual machine workload prediction method, a virtual machine placement method, and a virtual machine placement apparatus implementing the same, which can manage data stably by reducing the number of migrations in order to solve the above problems would like to provide

However, the problems to be solved by the present invention are not limited to the above-mentioned problems, and the problems not mentioned can be clearly understood by those of ordinary skill in the art to which the present invention belongs from the present specification and accompanying drawings. There will be.

delete

delete

delete

delete

delete

delete

delete

delete

delete

delete

delete

delete

delete

delete

delete

A virtual machine placement method according to an embodiment of the present invention includes a physical server in which a virtual machine is installed using a virtual machine placement device, wherein the physical server includes a first physical server and a second physical server. In the virtual machine arrangement method, which is a method of disposing a plurality of the virtual machines on the first physical server and the second physical server in order to efficiently operate the calculating virtual machine workload information that is information related to a virtual machine workload that is a load of a physical server imposed by the virtual machine by a predetermined workload calculation method based on operation-related information; calculating initial predicted virtual load information, which is information related to an initial predicted virtual load associated with a load of a physical server imposed by the virtual machine, predicted for a first predetermined period based on the virtual machine workload; calculating arrangement schedule information, which is information related to arrangement schedules of the virtual machines on the first physical server and the second physical server, by a predetermined arrangement schedule calculating method; displaying the arrangement schedule calculated by the predetermined arrangement schedule calculation method; and the virtual machines are arranged between the first physical server and the second so that the virtual machines are placed on the first physical server and the second physical server according to the arrangement schedule calculated by the predetermined arrangement schedule calculation method. and disposing while moving on a physical server, wherein the method for calculating the predetermined arrangement schedule includes calculating the arrangement schedule of the virtual machine when an objective function value is minimized based on the initial predicted virtual load information. and wherein the arrangement schedule is calculated so that the movement of the virtual machine between the physical servers is allowed during the first predetermined period, wherein the objective function is, during the first predetermined period, the initial predicted virtual load. A first objective function, which is a function related to the amount of change in the load of the physical server calculated by
In addition, the predetermined arrangement calculation method may be a method in which the arrangement schedule is calculated so that movement of the virtual machine between physical servers is allowed within the predetermined number of times during the first predetermined period.
Also, the predetermined workload calculation method may be a calculation method assuming that the specifications of the first physical server and the second physical server are the same.
Also, the predetermined workload calculation method may include a method of calculating the virtual machine workload information based only on the operation degree of the CPU.
In addition, the objective function includes a second objective function that is a function related to a difference between loads of a physical server calculated by the initial predicted virtual load during the first predetermined period, and the virtual machine is deployed according to the deployment schedule. a third objective function, which is a function related to the number of times the virtual machines are batched and moved on the first physical server and the second physical server to become 2 It may further include a fourth objective function that is a function related to whether the physical server is operating.
Also, in the arrangement schedule, the first physical server in which the values of the second objective function, the third objective function, and the fourth objective function are minimized based on a state in which the value of the first objective function is the minimum. and a first arrangement schedule that is the arrangement of the virtual machine on the second physical server, a value of the first objective function, a value of the third objective function, and a fourth objective based on a state in which the value of the second objective function is a minimum Based on the second arrangement schedule, which is the arrangement of the virtual machines on the first physical server and the second physical server, in which the value of the function is minimized, and the state in which the value of the third objective function is minimum, value, a third deployment schedule that is the placement of the virtual machines on the first physical server and the second physical server at which the values of the second objective function and the fourth objective function are minimized, and the value of the fourth objective function is the minimum Based on a state, a fourth deployment schedule is the arrangement of the virtual machines on the first physical server and the second physical server in which the values of the first objective function, the second objective function, and the third objective function are minimized. can be provided.
Also, the virtual machine may be deployed on the first physical server and the second physical server according to one of the deployment schedules selected from among the deployment schedules.
A virtual machine arrangement apparatus according to an embodiment of the present invention includes a physical server in which a virtual machine is installed, wherein the physical server includes a first physical server and a second physical server. - A virtual machine arrangement apparatus for implementing a virtual machine arrangement method, which is a method of disposing a plurality of the virtual machines on the first physical server and the second physical server in order to efficiently operate: a processing unit; and a storage unit storing a program for performing the virtual machine arrangement method, wherein the virtual machine arrangement method calculates a predetermined workload based on information related to operations of the first physical server and the second physical server calculating virtual machine workload information that is information related to a virtual machine workload that is a load of a physical server imposed by the virtual machine by the method; calculating initial predicted virtual load information that is information related to an initial predicted virtual load associated with a load of a physical server imposed by the virtual machine predicted for a first predetermined period based on the virtual machine workload; calculating arrangement schedule information, which is information related to arrangement schedules of the virtual machines on the first physical server and the second physical server, by a predetermined arrangement schedule calculating method; displaying the arrangement schedule calculated by the predetermined arrangement schedule calculation method; and the virtual machines are arranged between the first physical server and the second so that the virtual machines are placed on the first physical server and the second physical server according to the arrangement schedule calculated by the predetermined arrangement schedule calculation method. and disposing while moving on a physical server, wherein the method for calculating the predetermined arrangement schedule includes calculating the arrangement schedule of the virtual machine when an objective function value is minimized based on the initial predicted virtual load information. and wherein the arrangement schedule is calculated so that the movement of the virtual machine between the physical servers is allowed during the first predetermined period, wherein the objective function is, during the first predetermined period, the initial predicted virtual load. It may include a first objective function, which is a function related to the amount of change in the load of the physical server calculated by the
In addition, the predetermined arrangement calculation method may be a method in which the arrangement schedule is calculated so that movement of the virtual machine between physical servers is allowed within the predetermined number of times during the first predetermined period.
Also, the predetermined workload calculation method may be a calculation method on the assumption that the specifications of the first physical server and the second physical server are the same.
In addition, the objective function includes a second objective function that is a function related to a difference between loads of a physical server calculated by the initial predicted virtual load during the first predetermined period, and the virtual machine is deployed according to the deployment schedule. a third objective function, which is a function related to the number of times the virtual machines are batched and moved on the first physical server and the second physical server, to be a third objective function, whether the first physical server is operated during the predetermined first period, and the first 2 It may further include a fourth objective function, which is a function related to whether or not the physical server is operated.

A virtual machine workload prediction method, a virtual machine placement method, and a virtual machine placement apparatus implementing the same according to the present invention can minimize data loss.

In addition, resource consumption can be minimized.

In addition, the data center can be managed stably.

However, the effects of the present invention are not limited to the above-described effects, and the effects not mentioned will be clearly understood by those of ordinary skill in the art to which the present invention belongs from the present specification and accompanying drawings.

1 is a conceptual diagram of a virtual machine arrangement apparatus according to an embodiment of the present invention and a diagram illustrating a relationship between the virtual machine arrangement apparatus and a plurality of physical servers;
2 is a flowchart of a virtual machine placement method according to an embodiment of the present invention;
3 is a graph for explaining an effect achieved as a virtual machine arrangement in consideration of a first objective function in a virtual machine arrangement method according to an embodiment of the present invention;
4 is a graph for explaining an effect achieved as a virtual machine arrangement in consideration of a second objective function in a virtual machine arrangement method according to an embodiment of the present invention;
5 is a diagram for explaining a step of calculating a batch schedule in a virtual machine arrangement method according to an embodiment of the present invention;
6 is a graph for explaining a method of calculating a prediction error in a virtual machine arrangement method according to an embodiment of the present invention;

Hereinafter, specific embodiments of the present invention will be described in detail with reference to the drawings. However, the spirit of the present invention is not limited to the embodiments presented, and those skilled in the art who understand the spirit of the present invention may add, change, delete, etc. other components within the scope of the same spirit, through addition, change, deletion, etc. Other embodiments included within the scope of the invention may be easily proposed, but this will also be included within the scope of the invention.

In addition, components having the same function within the scope of the same idea shown in the drawings of each embodiment will be described using the same reference numerals.

In the present specification, when it is determined that a detailed description of a known configuration or function related to the present invention may obscure the gist of the present invention, a detailed description thereof will be omitted.

1 is a conceptual diagram of a virtual machine arrangement apparatus and a diagram illustrating a relationship between the virtual machine arrangement apparatus and a plurality of physical servers according to an embodiment of the present invention.

Referring to FIG. 1 , a virtual machine arrangement apparatus 10 according to an embodiment of the present invention arranges a processing unit 11 that processes information (eg, data) necessary for placing and arranging a virtual machine, a virtual machine, and It may include a storage unit 12 for storing information necessary for sorting, a receiving unit 13 for receiving information necessary for arranging and aligning virtual machines, and a transmitter 14 for transmitting information necessary for arranging and aligning virtual machines. have.

In addition, the virtual machine arrangement device 10 displays an input unit 15 that transmits information generated by external input to the receiving unit 13 and an arrangement schedule to be described later to deliver the arrangement schedule to the user. The medium may further include a display unit 16 .

The processing unit 11 may perform an operation such as processing all information necessary to implement the virtual machine arrangement method based on the information stored in the storage unit.

For example, the processing unit 11 may mean a central processing unit (CPU).

However, the present invention is not limited thereto, and the processing unit 11 may include any device or component capable of performing an operation required to implement the virtual machine arrangement method.

The storage unit 12 may store information (data) necessary for implementing the virtual machine arrangement method.

A program for performing the virtual machine arrangement method may be stored in the storage unit 12 .

For example, the storage unit 12 may be a hard disk as an auxiliary storage device.

However, the present invention is not limited thereto, and the storage unit 12 may include any element capable of storing information necessary for implementing the virtual machine arrangement method.

For example, the storage unit 12 may be a database.

The receiving unit 13 may receive information transmitted from a physical server.

The receiving unit 13 may receive information related to the operation of the physical server from a corresponding physical server.

The transmitter 14 may transmit information to the physical server.

The transmitter 14 may transmit the control signal calculated by the processor 11 to the physical server.

For this reason, the virtual machine arrangement device 10 may transmit a control signal to each physical server so that a virtual machine driven on one physical server is migrated to another physical server.

For example, the input unit 15 may be a keyboard.

However, without being limited thereto, the type of the input unit 15 may be variously modified at a level that is obvious to those skilled in the art.

For example, the display unit 16 may be a display.

However, the present invention is not limited thereto, and the type of the display unit 16 may be variously modified at a level apparent to those skilled in the art.

The components in the virtual machine placement device 10 may be connected to each other by wire and/or wirelessly.

The virtual machine arrangement device 10 may be connected to each of the physical servers by wire and/or wirelessly.

The physical server may refer to a hardware platform on which a virtual machine may be executed.

As an example, the physical server may be a typical desktop, server computer, laptop computer, or mobile phone or the like.

However, without being limited thereto, the physical server may be a concept including all means by which a virtual machine can be executed.

The virtual machine arrangement device 10 according to an embodiment of the present invention is described based on controlling and managing the first physical server 20, the second physical server 30, and the third physical server 40, The present invention is not limited thereto, and the number of physical servers managed and controlled by the virtual machine arrangement device 10 may be variously modified at a level apparent to those skilled in the art.

1 , the physical server includes a first physical server 20 , a second physical server 30 , and a third physical server 40 , and on the first physical server 20 , the a-th virtual The machine 21 , the b th virtual machine 22 , and the c th virtual machine 23 may be running, and on the second physical server 30 , the d th virtual machine 31 and the e th virtual machine 32 . ) may be being driven, and the f th virtual machine 41 and the g th virtual machine 42 may be driven on the third physical server 40 .

Hereinafter, a virtual machine arrangement method implemented using the virtual machine arrangement apparatus 10 will be described in detail.

2 is a flowchart of a virtual machine placement method according to an embodiment of the present invention.

Referring to FIG. 2 , in a virtual machine arrangement method according to an embodiment of the present invention, a physical server in which a virtual machine is installed, wherein the physical server includes a first physical server and a second physical server. In the virtual machine arrangement method, which is a method of disposing a plurality of the virtual machines on the first physical server and the second physical server in order to efficiently operate the A step of calculating virtual machine workload information that is information related to a virtual machine workload that is a load of a physical server imposed by the virtual machine by a predetermined workload calculation method based on operation-related information (S10), the virtual Calculating the initial predicted virtual load information, which is information related to the first predicted virtual load associated with the load of the physical server imposed by the virtual machine, predicted for a first predetermined period based on the machine workload (S20), a predetermined A step (S30) of calculating the arrangement schedule information, which is information related to the arrangement schedules of the virtual machines on the first physical server and the second physical server, by the arrangement schedule calculation method and the predetermined arrangement schedule calculation method A step of disposing the virtual machine while moving on the first physical server and the second physical server so that the virtual machine is placed on the first physical server and the second physical server according to the arrangement schedule (S50) may include

In addition, the virtual machine arrangement method may further include displaying a plurality of arrangement schedules calculated by the predetermined arrangement schedule calculation method by the display unit ( S40 ).

In addition, in the virtual machine arrangement method, a selected arrangement schedule, which is one arrangement schedule among the arrangement schedules displayed by the display unit, is input by the input unit, and information related to the selected arrangement schedule is transmitted to the processing unit (S50). may include more.

In addition, the virtual machine arrangement method is information related to a selection prediction virtual load associated with a load of a predicted physical server when the virtual machine is placed on the first physical server and the second physical server according to the selected placement schedule. A step (S70) of determining whether the selective prediction virtual load information is normally calculated according to a predetermined inspection method based on the selection prediction virtual load information and the actual virtual workload information that is information related to the actual workload of the virtual machine (S70) may include

The storage unit may store information related to the operation of the first physical server, information related to the operation of the second physical server, and information related to the operation of the third physical server.

For example, information related to operations of the first physical server, the second physical server, and the third physical server may be stored in the storage unit in time series.

To this end, the receiver may receive information related to each operation from the first physical server, the second physical server, and the third physical server.

For example, the information related to the operation may be information related to the utilization rate of the components of the physical server.

The processing unit is information related to a virtual machine workload that is a load of a physical server imposed by the virtual machine by a predetermined workload calculation method based on the information related to the operation of each physical server stored in the storage unit. Machine workload information may be calculated (S10).

When the virtual machine is driven on a physical server, the virtual machine workload may mean a load of resources of the corresponding physical server.

The virtual machine workload may mean a load applied to the physical server when the virtual machine is driven on the physical server.

For example, the virtual machine workload may be calculated as a load imposed on a central processing unit (CPU) of the physical server by the virtual machine being operated in the physical server.

Accordingly, the predetermined workload calculation method may include a method of calculating the virtual machine workload as a load imposed on a central processing unit of a physical server.

In other words, in the predetermined workload calculation method, the virtual machine workload information may be calculated based on only the operation degree of the CPU.

In the predetermined workload calculation method, it may be assumed that one virtual machine can be executed only on one physical server.

As an example, this may mean that one virtual machine cannot be implemented on two physical servers.

A workload of the physical server may mean a load imposed on the physical server.

For example, the workload of the physical server may be calculated as a load imposed on a central processing unit (CPU) of the physical server.

For example, the workload of the physical server may be calculated by the sum of virtual machine workloads that are loads (workloads) of virtual machines disposed on the physical server.

This simplifies the basic data required for prediction, which will be described later, thereby further increasing the accuracy of prediction in a step described later.

The processing unit is information related to an initial prediction virtual load associated with the load of the physical server imposed by the virtual machine that is predicted for a first predetermined period that is a future period from a reference point on the basis of the virtual machine workload information. Virtual load information may be calculated ( S20 ).

As an example, the reference time may be a virtual machine workload information calculation time.

However, without being limited thereto, the reference time point may be variously modified at a level obvious to those skilled in the art.

In order to calculate the initial predicted virtual load, virtual machine workload information that is a workload of each of the virtual machines during the first predetermined period may be predicted.

The reason for predicting the workload of the virtual machine for the first predetermined period and re-arranging the virtual machines based on this is, when migrating based on overload at a specific point in time, the number of migrations increases and the overall physical server stabilization because it cannot be implemented.

Accordingly, by disposing the virtual machine on the physical server based on a specific period rather than a specific point in the future, the above-described problem can be effectively solved.

As an example, the first predetermined period may be three months.

However, the present invention is not limited thereto, and the predetermined first period may be variously modified at a level obvious to those skilled in the art.

For example, the processing unit calculates virtual machine workload information for the first predetermined period by using an algorithm such as multiple regression model, ARIMA analysis, or deep learning based on the virtual machine workload information stored in the storage unit. can do.

The processing unit may calculate the predicted workload of the physical server for the first predetermined period based on the calculated virtual machine workload information for the first predetermined period.

This is a technique that is obvious to those of ordinary skill in the art, and a detailed description thereof may be omitted.

For example, the initial predicted virtual load may mean a workload of the virtual machine predicted for a first predetermined period from the time of calculating the virtual machine workload information or for a first predetermined period from an arbitrary time.

The processing unit may calculate ( S30 ) the arrangement schedule information, which is information related to the arrangement schedule, which is the arrangement of the virtual machines on the first physical server and the second physical server, by a predetermined arrangement schedule calculation method.

Here, the predetermined arrangement schedule calculation method may refer to a method of calculating the arrangement schedule of the virtual machine when an objective function value is minimized based on the initial predicted virtual load information.

The method of calculating the predetermined arrangement schedule may include a method of calculating the arrangement schedule of the virtual machine so that a load of the physical server is imposed on an arbitrary physical server only within a predetermined range. (1st condition)

For example, in the predetermined range, the operation rate (the degree to which a load is imposed) of the physical server may be 10% to 80%.

However, without being limited thereto, the predetermined range may be variously modified at a level apparent to those skilled in the art.

In the predetermined arrangement calculation method, the load of the arbitrary physical server is set according to a first criterion (X10, see FIGS. 3 and 4) so that the load of the physical server can be imposed on the arbitrary physical server only within a predetermined range. ) is predicted to be abnormal, the method may include a method of moving at least one virtual machine among the virtual machines disposed in the arbitrary physical server to another physical server. (Condition 1-1)

For example, the first criterion (X10, see FIGS. 3 and 4 ) may mean that an operation rate (a degree to which a load is imposed) of the physical server is 80%.

The processing unit is a method for preventing the load of the physical server from exceeding a first criterion, and based on a point in time when the workload of the target physical server is predicted to exceed the first criterion, at least one The deployment schedule may be calculated using a method of moving a virtual machine to another physical server.

For this reason, it is possible to effectively prevent the workload (operability rate) of the physical server from being excessively applied beyond the first criterion.

However, the present invention is not limited thereto, and the first standard may be variously modified at a level obvious to those skilled in the art.

In addition, in the predetermined arrangement calculation method, the load of the arbitrary physical server is set according to the second criterion (X20, FIGS. 3 and 3 and FIG. 4) and below, the method may include a method of moving all of the virtual machines disposed in the arbitrary physical server to another physical server. (Condition 1-2)

For example, the second criterion (X20, refer to FIGS. 3 and 4 ) may mean that an operation rate (a degree to which a load is imposed) of the physical server is 10%.

The processing unit is a solution to prevent the load of the physical server from being less than or equal to the second reference, and based on a point in time when the workload of the target physical server is predicted to fall below the second reference, all virtual machines placed on the target physical server A method of moving machines to another physical server can be used to calculate the deployment schedule.

This may mean an initiation condition of a condition in which the physical server is not operated.

In other words, the processing unit may stop the operation of the target physical server by disposing a virtual machine disposed in the target physical server to another physical server as a solution to prevent the load of the physical server from being lower than the second reference.

The predetermined arrangement calculation method may include a method in which the arrangement schedule information is calculated so that the physical server in operation does not change to a non-operational state due to movement of the virtual machine that causes the inactive physical server to operate. can (Second condition)

When stopping one physical server and starting the operation of another physical server at the same time, there is no benefit to stopping the physical server, and there is a problem in that more power and time are required to start the server.

Accordingly, through the second condition, such a problem may be solved.

A detailed description thereof will be given later.

In addition, when the batch schedule is calculated using the predetermined arrangement calculation method, the arrangement schedule may be calculated so that movement of the virtual machine between the physical servers is allowed during the first predetermined period. (3rd condition)

However, since data loss may be induced when the virtual machine is moved frequently, the movement of one virtual machine during the first predetermined period may be limited to a predetermined number of times. (4th condition)

For example, the predetermined number of times may be two times.

However, without being limited thereto, the predetermined number of times may be variously modified at a level obvious to those skilled in the art.

In addition, when the batch schedule is calculated using the predetermined batch calculation method, it is possible to calculate the batch schedule assuming that the specifications of each physical server are the same (homogenous condition). (5th condition)

The objective function may include a first objective function, a second objective function, a third objective function, and a fourth objective function.

The processing unit may calculate four non-dominant values of each objective function having a minimum value.

Accordingly, according to the predetermined arrangement schedule calculation method, four arrangement schedules can be calculated.

Hereinafter, the method for calculating the predetermined batch schedule will be described in detail.

3 is a graph for explaining an effect achieved as a virtual machine arrangement in consideration of a first objective function in a virtual machine arrangement method according to an embodiment of the present invention.

Referring to FIG. 3 , the objective function may include a first objective function that is a function related to the amount of change in the load (workload) of the physical server calculated by the first predicted virtual load during the first predetermined period. have.

For example, FIG. 3A is a graph illustrating workloads of the first to third physical servers when virtual machines are disposed in first to third physical servers according to a first schedule.

Also, FIG. 3B is a graph illustrating workloads of the first to third physical servers when virtual machines are disposed in the first to third physical servers according to the second schedule.

It can be seen from the above description that the workload of each physical server can be calculated as the sum of the virtual machine workloads installed in each physical server.

Referring to FIG. 3A , when virtual machines are disposed on a physical server according to a first schedule, the workload of all physical servers may be less than the first criterion (X10) and greater than the second criterion (X20). have.

In addition, referring to FIG. 3( b ), when virtual machines are disposed on a physical server according to the second schedule, the workload of all physical servers is less than the first criterion (X10) and the second criterion is exceeded (X20). can be

However, when the virtual machines are arranged on the physical server based on the second schedule, the amount of change (slope) of the workload of each physical server is higher than when the virtual machines are arranged on the physical server based on the first schedule. can be smaller

This may be a schedule in which the second schedule is more preferable than the first schedule as the deployment schedule of the virtual machine on the physical server.

In addition, a value of the first objective function from which the second schedule is derived may be smaller than a value of the first objective function from which the first schedule is derived.

If the change rate of the workload of the physical server is large, the workload of the physical server exceeds an arbitrary criterion only for a short period of time, and migration (transfer) of the virtual machine can be made.

In this case, since the above-described frequent migration of the virtual machine may occur, the risk of data loss of the virtual machine may further increase.

The first objective function may be a function for preventing this.

4 is a graph for explaining an effect achieved as a virtual machine arrangement in consideration of a second objective function in a virtual machine arrangement method according to an embodiment of the present invention.

Referring to FIG. 4 , the objective function may include a second objective function that is a function related to a difference between loads (workloads) of a physical server calculated by the initial predicted virtual load during the first predetermined period. have.

For example, FIG. 4A is a graph illustrating workloads of the first to third physical servers when virtual machines are disposed on the first to third physical servers according to a third schedule.

Also, FIG. 4B is a graph illustrating workloads of the first to third physical servers when virtual machines are disposed in the first to third physical servers according to the fourth schedule.

It can be seen from the above description that the workload of each physical server can be calculated as the sum of the virtual machine workloads installed in each physical server.

Referring to FIG. 4(a) , when virtual machines are placed on a physical server according to the third schedule, the workload of all physical servers may be less than the first criterion (X10) and greater than the second criterion (X20). have.

In addition, referring to FIG. 4( b ), when virtual machines are disposed on a physical server according to the fourth schedule, the workload of all physical servers is less than the first criterion (X10) and the second criterion (X20) is greater than can be

However, when the virtual machines are arranged on the physical server based on the fourth schedule, the difference between the workloads of each physical server is smaller than when the virtual machines are arranged on the physical server based on the third schedule. can

This may be a schedule in which the fourth schedule is more preferable than the third schedule as the deployment schedule of the virtual machine on the physical server.

Also, the value of the second objective function for deriving the fourth schedule may be smaller than the value of the second objective function for deriving the third schedule.

Here, in order to compare the workloads between the physical servers, the physical servers may be compared with each other by using an average value of the workloads of the physical servers during the first predetermined period.

The meaning of the second objective function may be to evenly utilize the physical server.

The objective function may include a third objective function, which is a function related to the number of times the virtual machine is placed and moved on the first physical server and the second physical server so that the virtual machine is placed according to the placement schedule.

In other words, the third objective function may be related to the number of times the virtual machine moves between physical servers to be deployed according to a deployment schedule in the current deployment of virtual machines.

The meaning of the third objective function may be to find an arrangement that minimizes the number of movements of the virtual machine.

Whenever a virtual machine is migrated, there is always a risk of data loss.

The meaning of the third objective function may be to minimize the possibility of data loss.

The objective function may include a fourth objective function that is a function related to whether the first physical server is operated and whether the second physical server is operated during the first predetermined period.

In other words, the fourth objective function may be a function related to whether the physical server is operated during the first predetermined period.

As the fourth objective function is smaller, the number of physical servers that are not in operation and the time the physical servers are not in operation can be maximized.

When the fourth objective function is the minimum, the sum of idle times per one physical server for all physical servers may be maximized.

The meaning of the fourth objective function may be meaningful in minimizing the power loss caused by the operation of the physical server by minimizing the server operation time.

The arrangement schedule may mean a schedule for how virtual machines are arranged on physical servers so that each objective function can be minimized in a state where the first to fifth conditions are satisfied.

The objective function includes a first objective function that is a function related to an amount of change in a load of a physical server calculated by the initial predicted virtual load during the first predetermined period, and during the first predetermined period, the first predicted virtual load A second objective function, which is a function related to a difference between loads of a physical server calculated by A third objective function, which is a function related to the number of times, and a fourth objective function, which is a function related to whether or not the first physical server is operated and whether the second physical server is operated during the first predetermined period, may be provided.

The display unit may display the batch schedules transmitted from the processing unit ( S40 ).

The processing unit includes first batch schedule information, which is information related to the first batch schedule, second batch schedule information, which is information related to the second batch schedule, third batch schedule information, and fourth batch schedule information, which is information related to a third batch schedule. After calculating the fourth arrangement schedule information, which is information related to , it may be transmitted to the display unit.

The arrangement schedule includes physical servers in which the values of the second objective function, the third objective function, and the fourth objective function are minimized based on the state in which the value of the first objective function is the minimum (eg, the second objective function). A value of the first objective function and a value of the third objective function based on a first configuration schedule that is the configuration of the virtual machine on the first physical server and the second physical server) and a state in which the value of the second objective function is a minimum and a second arrangement schedule that is the arrangement of the virtual machine on physical servers (eg, a first physical server and a second physical server) in which the value of the fourth objective function is minimized, and a state in which the value of the third objective function is the minimum. placement of the virtual machine on physical servers (eg, the first physical server and the second physical server) in which the values of the first objective function, the second objective function, and the fourth objective function are minimized based on Physical servers in which the values of the first objective function, the second objective function, and the third objective function are minimized based on a third batch schedule of , the first physical server and the second physical server) may include a fourth deployment schedule that is the deployment of the virtual machine.

Hereinafter, a method of calculating a predetermined arrangement schedule will be described using one arrangement schedule as an example.

FIG. 5 is a diagram for explaining a step of calculating a batch schedule in a virtual machine arrangement method according to an embodiment of the present invention.

FIG. 5 is a diagram schematically illustrating schedules of virtual machines disposed in physical servers for a predetermined period among a first predetermined period.

As an example, the example shown in FIG. 5 may relate to a first batch schedule based on a first objective function being minimized, and hereinafter, it will be described based on this.

However, it is obvious that this is an example and the present invention is not limited thereto.

For example, the physical server may include a first physical server 20 , a second physical server 30 , and a third physical server 40 , and the virtual machine includes a first virtual machine 91 , a second virtual machine ( 92 , a third virtual machine 93 , a fourth virtual machine 94 , a fifth virtual machine 95 , a sixth virtual machine 96 , and a seventh virtual machine 97 .

For example, the processing unit includes the first virtual machine 91 and the second virtual machine 91 in the first physical server 20 at time 0, which is a time before the first time, so that other objective functions can be minimized while the first objective function is minimized. The second virtual machine 92 is placed, and the third virtual machine 93 , the fourth virtual machine 94 , and the fifth virtual machine 95 are placed on the second physical server 30 , and the third physical The server 40 may calculate a deployment schedule so that the sixth virtual machine 96 and the seventh virtual machine 97 are placed.

If the placement of virtual machines in the 0 th time zone is followed, it may be predicted that the workload of the first physical server is less than or equal to the second reference in the first time period.

In this case, as a case in which condition 1-2 is satisfied, the first virtual machine 91 and the second virtual machine 92, which are virtual machines arranged in the first physical server 20, are moved to different physical servers. A batch schedule may be calculated as much as possible.

Here, when the virtual machine is transferred under the condition 1-2, it may be determined to which physical server the virtual machine is transferred based on the minimized first objective function.

Based on this, the arrangement schedule may be calculated such that the first virtual machine 91 is moved to the second physical server 30 and the second virtual machine 92 is moved to the third physical server 40 .

By the migration at the first time, the arrangement schedule may be calculated so that the first physical server 20 is not operated at the second time.

For example, the third virtual server 30 , the fourth virtual server 40 , the fifth virtual server 50 , and the first virtual server 10 are disposed in the second physical server 30 , and the third physical server When the second virtual server 92 , the sixth virtual server 96 , and the seventh virtual server 97 are arranged in 40 , it is assumed that the other objective functions are also minimum while the first objective function is minimum. , the migration may not occur at the second time.

On the other hand, in order to minimize the other objective functions while the first objective function is the minimum, the first virtual server 91 disposed in the second physical server 30 at the third time is performed by the third physical server 40 . ) can be calculated to be migrated to

In addition, in order to minimize the other objective functions while the first objective function is the minimum, the fifth virtual server 95 arranged in the second physical server 30 at the fourth time is performed by the third physical server 40 . A deployment schedule can be calculated to migrate to

The first physical server 20 may not be operated during the second to fourth hours.

This may be that the arrangement schedule is calculated so that the first physical server 20 is not operated for the second to fourth time due to the influence of the fourth objective function.

If the arrangement of the virtual machines at the fourth time is followed, it may be predicted that the workload of the third physical server 40 exceeds the first criterion at the fifth time.

In this case, the arrangement schedule may be calculated such that at least one of the virtual machines arranged in the third physical server 40 is moved to another physical server at the fifth time. (Condition 1-1)

In addition, if the arrangement of the virtual machines at the fourth time is followed, it may be predicted that the workload of the second physical server 30 exceeds the first criterion at the fifth time. (Condition 1-1)

In this case, the arrangement schedule may be calculated so that at least one of the virtual machines arranged in the second physical server 30 is moved to another physical server at the fifth time. (Condition 1-1)

However, since the first virtual machine 91 disposed in the third physical server 40 is migrated at the first time and the third time, it may be excluded from the migration target at the fifth time. (4th condition)

In addition, when both the third virtual machine 93 and the fourth virtual machine 94 arranged in the second physical server 30 are moved to the first physical server 20, the second physical server 30 is operated. is stopped and the operation of the first physical server 20 may be started.

Since the second condition is not satisfied, in the second physical server 30 , only one virtual machine among the third virtual machine 93 and the fourth virtual machine 94 can be moved to the first physical server 20 . have.

For example, in a state in which the first objective function is minimized, all objective functions are minimized, and the first to fourth conditions (specifically, the 1-2 conditions and the second conditions) are satisfied, the second physical server 30 In , the third virtual machine 93 may be moved to the first physical server 20 , and in the third physical server 40 , the sixth virtual machine 96 and the seventh virtual machine 97 are the first physical server (20) can be moved.

A batch schedule may be calculated so that the minimum value of the objective function is satisfied by the migration at the fifth time and migration does not occur at the sixth time.

So far, the description has been made based on the first batch schedule, but this is only an example, and a detailed description of a method in which a schedule is calculated based on the second batch schedule, the third batch schedule, and the fourth batch schedule is described above. It may be omitted to the extent that it overlaps with one content.

The display unit may display the first arrangement schedule, the second arrangement schedule, the third arrangement schedule, and the fourth arrangement schedule.

Since the four objective functions can each have four non-dominant solutions as independent functions, the first batch schedule, the second batch schedule, the third batch schedule, and the fourth batch schedule are independent of each other. can mean

The operator may determine one of the first batch schedule, the second batch schedule, the third batch schedule, and the fourth batch schedule through the input unit, and select and input the determined batch schedule.

The input unit may calculate information related to the selected arrangement schedule and transmit the information to the processing unit.

6 is a graph for explaining a method of calculating a prediction error in a virtual machine arrangement method according to an embodiment of the present invention.

The processing unit may deploy (S60) virtual machines while moving them on the physical servers based on the selected deployment schedule.

Through this method, the resources of the physical server can be used evenly without being biased toward one physical server.

Also, in the virtual machine arrangement method according to an embodiment of the present invention, when the virtual machines are placed in the first physical server and the second physical server according to the arrangement schedule, selection associated with the predicted load of the physical server It is determined whether the selection prediction virtual load information is normally calculated according to a predetermined inspection method based on the selection prediction virtual load information, which is information related to the prediction virtual load, and the actual virtual workload information, which is information related to the actual workload of the virtual machine. The step of determining (S70) may be further included.

Here, the selective prediction virtual load information may be the same as the above-described initial prediction virtual load information.

However, the present invention is not limited thereto, and the selection prediction virtual load information may not match the initial prediction virtual load information.

In this case, after arranging the virtual machines according to the selected arrangement schedule, the processing unit may calculate the selected predicted virtual load information by predicting the workload of the virtual machines for a predetermined period.

The predetermined checking method may be a method of determining that the selective prediction virtual load is erroneously calculated when the actual workload of the virtual machine differs from the corresponding selective prediction virtual load by a predetermined value or more.

As a specific example, FIG. 6 shows a first workload Y10 representing a selective prediction virtual load that is a predicted workload of the a-th virtual machine and a second workload Y20 representing the actual workload of the a-th virtual machine. It can be a graph.

The processing unit may determine that the workload of the a-th virtual machine is erroneously predicted when the first workload and the second workload differ by more than a predetermined workload (a predetermined value, Y).

When the processing unit determines that the selection prediction virtual load is calculated incorrectly, the processing unit determines that the selection prediction virtual load is calculated incorrectly, for a second predetermined period from the time of re-judgment, by the predicted virtual machine. Re-prediction virtual load information, which is information related to the re-prediction virtual load associated with the generated load of the physical server, may be calculated ( S80 ).

The processing unit may re-calculate (S80) the batch schedule by the above-described predetermined batch schedule calculation method based on the re-prediction virtual load information, and based on this, the virtual machines can be re-arranged (S90). can be calculated.

A detailed description thereof may be omitted to the extent that it overlaps with the above-described content.

Here, the second predetermined period may be the same as or different from the first predetermined period described above.

The virtual machine arrangement method as described above may be repeatedly and continuously performed.

In order for the virtual machines to be relocated (S90), an operator may select one arrangement schedule from among a plurality of arrangement schedules, and the processing unit may rearrange the virtual machines based on the selected arrangement schedule (S90).

A detailed description thereof may be omitted to the extent that it overlaps with the above-described content.

A virtual machine arrangement apparatus according to an embodiment of the present invention includes a physical server in which a virtual machine is installed, wherein the physical server includes a first physical server and a second physical server. - A virtual machine arrangement apparatus for implementing a virtual machine arrangement method, which is a method of disposing a plurality of the virtual machines on the first physical server and the second physical server in order to efficiently operate: a processing unit; and a storage unit storing a program for performing the virtual machine arrangement method, wherein the virtual machine arrangement method calculates a predetermined workload based on information related to operations of the first physical server and the second physical server calculating virtual machine workload information that is information related to a virtual machine workload that is a load of a physical server imposed by the virtual machine by a method, the virtual machine workload information predicted for a first predetermined period based on the virtual machine workload Calculating initial predicted virtual load information, which is information related to an initial predicted virtual load associated with a load of a physical server imposed by a virtual machine, on the first physical server and the second physical server by a predetermined arrangement schedule calculation method Calculating arrangement schedule information, which is information related to the arrangement schedule of the virtual machine, and the virtual machine on the first physical server and the second physical server according to the arrangement schedule calculated by the predetermined arrangement schedule calculation method and disposing the virtual machine while moving on the first physical server and the second physical server, so that the predetermined arrangement schedule calculation method includes an objective function value based on the initial predicted virtual load information. This may be a method in which the deployment schedule of the virtual machine when this is minimized is calculated.

In the accompanying drawings, in order to more clearly express the technical spirit of the present invention, components that are not related to or inferior to the technical spirit of the present invention are briefly expressed or omitted.

In the above, the configuration and features of the present invention have been described based on the embodiments according to the present invention, but the present invention is not limited thereto, and it is understood that various changes or modifications can be made within the spirit and scope of the present invention. It is intended that such changes or modifications will be apparent to those skilled in the art, and therefore fall within the scope of the appended claims.

10: virtual machine placement device 11: processing unit
12: storage unit 13: receiving unit
14: transmitter 15: input unit
16: display unit

Claims (11)

A physical server in which a virtual machine is installed using a virtual machine arrangement device, wherein the physical server includes a first physical server and a second physical server. In the virtual machine arrangement method, which is a method of disposing a plurality of the virtual machines on the first physical server and the second physical server in order to efficiently operate the
A virtual machine work that is information related to a virtual machine workload that is a load of a physical server imposed by the virtual machine by a predetermined workload calculation method based on information related to the operation of the first physical server and the second physical server calculating load information;
calculating initial predicted virtual load information, which is information related to an initial predicted virtual load associated with a load of a physical server imposed by the virtual machine, predicted for a first predetermined period based on the virtual machine workload;
calculating arrangement schedule information, which is information related to arrangement schedules of the virtual machines on the first physical server and the second physical server, by a predetermined arrangement schedule calculating method;
displaying the arrangement schedule calculated by the predetermined arrangement schedule calculation method; and
The virtual machine is arranged between the first physical server and the second physical server so that the virtual machine is disposed on the first physical server and the second physical server according to the arrangement schedule calculated by the predetermined arrangement schedule calculation method. Including; the step of being placed while moving on the server;
The predetermined batch schedule calculation method is,
A method for calculating the deployment schedule of the virtual machine when an objective function value is minimized based on the initial predicted virtual load information,
a method in which the arrangement schedule is calculated so that the movement of the virtual machine between the physical servers is allowed during the first predetermined period,
The objective function is
and a first objective function that is a function related to the amount of change in the load of the physical server calculated by the first predicted virtual load during the first predetermined period,
How to deploy virtual machines.
According to claim 1,
The predetermined batch calculation method is,
a method in which the deployment schedule is calculated such that movement of a virtual machine between physical servers is allowed within the first predetermined period within a predetermined number of times,
How to deploy virtual machines.
According to claim 1,
The predetermined workload calculation method,
It is a method of calculating assuming that the specifications of the first physical server and the second physical server are the same,
How to deploy virtual machines.
According to claim 1,
The predetermined workload calculation method,
Including a method for calculating the virtual machine workload information based only on the operation degree of the CPU,
How to deploy virtual machines.
According to claim 1,
The objective function is
During the first predetermined period, a second objective function that is a function related to a difference between loads of a physical server calculated by the initial predicted virtual load, the first physical server for the virtual machine to be deployed according to the deployment schedule and a third objective function, which is a function related to the number of times the virtual machine is placed and moved on the second physical server, whether the first physical server is operated and whether the second physical server is operated during the first predetermined period; Further comprising a fourth objective function that is a related function,
How to deploy virtual machines.
6. The method of claim 5,
The deployment schedule is
Based on the state in which the value of the first objective function is minimum, the values of the second objective function, the third objective function, and the fourth objective function are minimized on the first physical server and the second physical server a first deployment schedule that is a deployment of the virtual machine;
Based on the state in which the value of the second objective function is minimum, the values of the first objective function, the third objective function, and the fourth objective function are minimized on the first physical server and the second physical server a second deployment schedule that is the deployment of the virtual machine;
Based on the state in which the value of the third objective function is the minimum, the values of the first objective function, the second objective function, and the fourth objective function are minimized on the first physical server and the second physical server. a third deployment schedule that is the deployment of the virtual machine; and
Based on the state in which the value of the fourth objective function is the minimum, the values of the first objective function, the second objective function, and the third objective function are minimized on the first physical server and the second physical server. having a fourth deployment schedule that is the deployment of the virtual machine,
How to deploy virtual machines.
7. The method of claim 6,
The virtual machine is
deployed on the first physical server and the second physical server according to the one selected from among the deployment schedules;
How to deploy virtual machines.
A physical server in which a virtual machine is installed - The physical server includes a first physical server and a second physical server. - in a virtual machine arrangement apparatus implementing a virtual machine arrangement method, which is a method of disposing a plurality of the virtual machines on the first physical server and the second physical server in order to efficiently operate
processing unit; and
and a storage unit in which a program for performing the virtual machine arrangement method is stored;
The virtual machine deployment method comprises:
A virtual machine work that is information related to a virtual machine workload that is a load of a physical server imposed by the virtual machine by a predetermined workload calculation method based on information related to the operation of the first physical server and the second physical server calculating load information;
calculating initial predicted virtual load information, which is information related to an initial predicted virtual load associated with a load of a physical server imposed by the virtual machine, predicted for a first predetermined period based on the virtual machine workload;
calculating arrangement schedule information, which is information related to arrangement schedules of the virtual machines on the first physical server and the second physical server, by a predetermined arrangement schedule calculating method;
displaying the arrangement schedule calculated by the predetermined arrangement schedule calculation method; and
The virtual machine is arranged between the first physical server and the second physical server so that the virtual machine is disposed on the first physical server and the second physical server according to the arrangement schedule calculated by the predetermined arrangement schedule calculation method. Including; the step of being placed while moving on the server;
The predetermined batch schedule calculation method is,
A method for calculating the deployment schedule of the virtual machine when an objective function value is minimized based on the initial predicted virtual load information,
a method in which the arrangement schedule is calculated so that the movement of the virtual machine between the physical servers is allowed during the first predetermined period,
The objective function is
and a first objective function that is a function related to the amount of change in the load of the physical server calculated by the first predicted virtual load during the first predetermined period,
virtual machine placement device.
9. The method of claim 8,
The predetermined batch calculation method is,
a method in which the deployment schedule is calculated such that movement of a virtual machine between physical servers is allowed within the first predetermined period within a predetermined number of times,
virtual machine placement device.
9. The method of claim 8,
The predetermined workload calculation method,
It is a method of calculating assuming that the specifications of the first physical server and the second physical server are the same,
virtual machine placement device.
9. The method of claim 8,
The objective function is
During the first predetermined period, a second objective function that is a function related to a difference between loads of a physical server calculated by the initial predicted virtual load, the first physical server for the virtual machine to be deployed according to the deployment schedule and a third objective function, which is a function related to the number of times the virtual machine is placed and moved on the second physical server, whether the first physical server is operated and whether the second physical server is operated during the first predetermined period; Further comprising a fourth objective function that is a related function,
virtual machine placement device.

Images