JP6331549B2 - Virtual machine management apparatus, virtual machine management method, and virtual machine management system - Google Patents

Description

  The present invention relates to a virtual machine management apparatus, a virtual machine management method, and a virtual machine management system, and in particular, a virtual machine management apparatus and a virtual machine management that select a virtual machine that is a movement candidate using time-series information of virtual machine performance values. The present invention relates to a method and a virtual machine management system.

  In recent years, the sophistication and complexity of the system has progressed, and the number of physical servers constituting the system has rapidly increased. At the same time, the labor and cost required for the maintenance and management of the server increase, which is a big problem. In addition, the server needs certain surplus resources to cope with peak loads. Server resources tend to swell because resources that are not normally used are also prepared. In a cloud environment, by virtualizing a server, a plurality of virtual machines can be physically built on one server and different systems can be operated. As a result, resource utilization efficiency increases dramatically.

  An example of a virtual machine migration management server that implements appropriate placement and migration of virtual machines using fluctuations in loads on virtual machines and physical machines is described in Patent Document 1.

  The virtual machine migration management server described in Patent Document 1 represents a change in the load of a virtual machine and a physical machine as a high-order equation, and a threshold value that is constant after adding the two high-order equations after normalization is set. It is determined that the load peaks of the two machines overlap with each other, and the physical machine with the smallest CPU load among the physical machines or the physical machine with no overlapping peaks is determined as the destination physical machine. Reduce the load.

  By the way, in recent cloud systems and the like, it is required to operate a large number of virtual machines on one physical machine. Many virtual machines share and consume physical machine resources, but for example, there are virtual machines whose resource consumption continues to fluctuate in response to processing requests. There is also a machine.

  However, as in the virtual machine migration management server described in Patent Document 1, the resource consumption can be identified by overlapping the load peak dynamically according to the processing request from the outside, and the resource consumption It is a virtual machine whose quantity has a tendency of time variation similar to that of a physical machine. By doing so, the resource consumption fluctuates dynamically, but the fluctuation range is relatively small relative to that of the physical machine, but the time variation of the resource consumption coincides with that of the physical machine. There is a problem in that the machine is specified by mistake, and as a result, the placement accuracy of the virtual machine deteriorates.

JP 2008-288593 A

  As in the virtual machine migration management technique described in Patent Document 1 above, the virtual to be moved based on the peak of the resource consumption when the resource consumption varies over time due to a fluctuating load on the virtual machine. When a machine is specified, there is a problem that an unintended virtual machine is specified as a virtual machine to be moved if there is a virtual machine whose resource consumption of the virtual machine is gradual.

  An object of the present invention is to solve the above-described problems and provide a virtual machine management device, a virtual machine management method, and a virtual machine management system that appropriately select a virtual machine to be moved.

  The virtual machine management apparatus of the present invention includes a physical unit having one or more virtual machines and a storage unit that stores performance information of the virtual machines in time series, and the influence of the performance information on the physical machines and the virtual machines. A selection unit that selects and outputs one migration candidate virtual machine based on the degree.

  The virtual machine management system of the present invention selects and outputs a physical machine having a plurality of virtual machines and a movement candidate virtual machine that is a candidate for moving from the physical machine to another physical machine among the plurality of virtual machines. A movement candidate machine selection device that moves the movement candidate virtual machine from the physical machine to another physical machine, and the movement candidate machine selection device includes the physical machine and the virtual machine. A storage unit that stores performance information in time series, a selection unit that selects and outputs one migration candidate virtual machine based on the influence degree of the performance information in the physical machine and each virtual machine, and the virtual And a virtual machine movement instruction unit that causes the movement candidate virtual machine to move to another physical machine.

  The virtual machine management method of the present invention stores a physical machine having one or more virtual machines and performance information of the virtual machine in time series, and is based on an influence degree of the performance information in the physical machine and each virtual machine. To select and output one migration candidate virtual machine.

  An effect of the present invention is that a virtual machine to be moved can be appropriately selected.

It is a block diagram which shows the characteristic structure of 1st embodiment of this invention. It is a flowchart which shows the whole process of the virtual machine management apparatus in 1st embodiment of this invention. It is a flowchart which shows the whole process of the virtual machine management apparatus in 1st embodiment of this invention. It is a block diagram which shows the structure of the virtual machine management system to which the virtual machine management apparatus in 1st embodiment of this invention is applied. It is a figure which shows the example of the performance information in 1st embodiment of this invention.

(First embodiment)
Next, a first embodiment of the present invention will be described.

  First, the configuration of the first embodiment of the present invention will be described. FIG. 4 is a block diagram showing a configuration of a performance information analysis system to which the virtual machine management apparatus according to the first embodiment of the present invention is applied.

  Referring to FIG. 4, the performance information analysis system according to the first exemplary embodiment of the present invention includes a virtual machine management device 1, a physical machine 2 (2-1, 2-2), and a virtual machine 3 (3-1, 3). -2), and the virtual machine moving device 4.

  Here, the virtual machine management device 1 selects the virtual machine 3 to be moved from the performance data acquired from the physical machine 2 and the virtual machine 3, and instructs the virtual machine moving device 4 to move based on the selection result. It is an apparatus for managing the virtual machine 3 to be performed.

  Here, the physical machine 2 is an apparatus that can execute one or more virtual machines 3. Each of the physical machine 2 and the virtual machine 3 performs processing according to a request, and internally generates the amount of resources used for the processing as performance data and can provide the same to the outside. In the example of FIG. 4, virtual machines 3-1 to 3-n are operating on the physical machine 2-1. The virtual machine is not operating on the physical machine 2-2.

  The virtual machine moving device 4 is a device for moving the virtual machine 3 so that the physical machine 2-1 that executes the virtual machine 3 is changed to another physical machine, for example, the physical machine 2-2.

  The virtual machine management device 1 includes a storage unit 11, a selection unit 12, an instruction unit 13, and a collection unit 14 (not shown).

  The storage unit 11 stores performance information such as CPU, memory, disk I / O, and network I / O collected from the physical machine 2 and the virtual machine 3. Specifically, data such as date / time, machine name, CPU usage rate, memory usage rate (usage amount), disk I / O, network I / O, and the like are accumulated.

The selection unit 12 selects the virtual machine 3 to be most moved from the data accumulated in the storage unit 11.
The instruction unit 13 receives from the selection unit 12 information on the virtual machine 3 that should be most moved, which is selected by the selection unit 12, and issues an instruction for movement.
The collection unit 14 collects performance information from the physical machine 2 and the virtual machine 3 and delivers the data to the storage unit 11.
FIG. 2 is a flowchart showing the overall processing of the virtual machine management device 1 according to the first embodiment of this invention.

  In step S101, the collection unit 14 determines the CPU, memory, disk I / O, network I / O, and the like of the physical machine 2-1 and the virtual machines 3-1 to 3-n arranged in the physical machine 2-1. Collect performance information. The performance information is collected in a time-series format and values collected at predetermined time intervals (for example, 1 minute).

  In step S102, the storage unit 11 stores the collected performance information. Typically, in a time series specified by a combination of a machine name and an item (CPU usage rate, memory usage rate, etc.), the value is stored in association with the time at which the value was collected.

  In step S103, the selection unit 12 acquires the performance information from the storage unit 11, and selects the virtual machine 3 to be moved by analyzing the acquired performance information. A specific method of selection will be described in detail in FIG. For example, it is assumed that the virtual machine 3-1 operating on the physical machine 2-1 is selected as a virtual machine to be moved.

  In step S <b> 104, the instruction unit 13 instructs the virtual machine moving device 4 to move the virtual machine based on the selected information on the virtual machine to be moved. The virtual machine moving device 4 performs a moving process so that the virtual machine 3-1 operating on the physical machine 2-1 operates on the physical machine 2-2, for example.

  Thus, a series of operations of the virtual machine management device 1 is completed.

  FIG. 3 is a flowchart showing a virtual machine selection process of the virtual machine management apparatus 1 according to the first embodiment of the present invention, and FIG. 5 is a physical machine 2 and a virtual machine according to the first embodiment of the present invention. It is a figure which shows the example of the data of 3 resources.

  In step S <b> 201, the selection unit 12 compares the performance information of the physical machine 2 with the threshold value, and extracts predetermined resource data. Here, the data is extracted by extracting data based on the machine name of the physical machine 2 for past data of a predetermined period such as one day or one month, and comparing the extracted data with a threshold value. Do. For comparison with the threshold value, for example, data of each time of the resource that is closest to the threshold value among the maximum values of the usage amount of each resource is specified and extracted. Other examples include, for example, the resource whose number of times the threshold has been exceeded among the maximum value of each resource usage, or the resource with the longest total time that exceeds the threshold among the maximum value of each resource usage. You may make it extract. For example, when the threshold is set to 30%, in FIG. 5, it is the CPU that exceeds 30%, so the CPU is specified and data is extracted.

  In step S <b> 202, the selection unit 12 extracts the resource data of each virtual machine 3 having resources related to the extracted data of the physical machine 2. In FIG. 5, since the virtual machine 3-1 and the virtual machine 3-2 have CPUs, these are extracted.

  In step S <b> 203, the selection unit 12 calculates the similarity between the physical machine 2 and each virtual machine 3 corresponding to the physical machine 2. The calculation of the similarity is a value obtained by accumulating the absolute values of the difference between adjacent values when the data of the specific resource in the physical machine 2 and the virtual machine 3 in a predetermined period is viewed in a time series format over the predetermined period. Ask for. That is, when the time is t, the specific resource data value of the physical machine 2 is P (t), and the specific resource data value of the virtual machine 3 is V (t), the similarity C is Σ ( t is expressed by | (P (t) −P (t−1)) − (V (t) −V (t−1)) |. This is performed for each of the virtual machines 3-1 to 3-n, and C1 to Cn are obtained. In FIG. 5, C in the virtual machine 3-1 is 0, and C in the virtual machine 3-2 is 50. Note that the predetermined period, which is the target period for calculating the amount of similarity, is, for example, any N hours (or one day, one week) regardless of the time, or any one day working hour (for example, from 8:00 to 17). Time). Further, intermittent time may be used instead of continuous time. For example, a weekday working hour in an arbitrary week may be set as the predetermined period. In this case, typically, data of five independent times may be obtained, and these may be treated as virtually continuous data to perform a required process. Since the predetermined period can be determined in various ways as described above, the similarity can be calculated efficiently or accurately according to the characteristics of the target business system.

  In step S204, the selection unit 12 calculates the influence level based on the similarity calculated in step S203. The influence degree is calculated from the similarity C for the virtual machines 3-1 to 3-n. The influence degree indicates how much the behavior of the virtual machine 3 affects the behavior of the resource usage of the physical machine 2 during a predetermined period. The influence degree is calculated based on C1 to Cn obtained in S203 so that the smaller the value, the larger the influence degree. As a result, a larger (higher) degree of influence is obtained for the virtual machine 3-1 having a smaller C.

  In step S205, the selection unit 12 compares the degree of influence of each virtual machine 3, and selects and determines the virtual machine 3 having the large degree of influence as a migration candidate virtual machine. In the example of FIG. 5, the virtual machine 3-1 is determined as a migration candidate virtual machine. The migration source of the migration candidate virtual machine is a physical machine on which the current virtual machine operates. The destination of the migration candidate virtual machine is another physical machine. The migration destination may be, for example, a physical machine having a resource usage lower than that of the migration source. As another example, the degree of influence is calculated between the virtual machine and the movement destination using a method for determining the migration candidate virtual machine, and the physical machine determined based on the calculated degree of influence, for example, the degree of influence is moved. It may be the physical machine that is lower or lowest than the original. Alternatively, the destination may be determined manually based on the resource usage amount. It is not always necessary to move a high-impact machine. The actual necessity of movement may be determined according to the situation. When it is desired to reduce the resource usage of a physical machine, leveling of the resource usage of the physical machine can be achieved by preferentially moving a virtual machine having a larger influence degree than others.

Next, a characteristic configuration of the first embodiment of the present invention will be described. FIG. 1 is a block diagram showing a characteristic configuration of the first embodiment of the present invention.

  Referring to FIG. 1, the virtual machine management apparatus according to the first embodiment of the present invention includes a storage unit 11 and a selection unit 12.

  Here, the storage unit 11 stores a plurality of pieces of performance information in the system in time series.

  The selection unit 12 evaluates the degree of influence from the performance information of different machines stored in the storage unit 11, and selects and outputs a movement candidate virtual machine that is the virtual machine 3 to be moved based on the evaluation.

  According to the first embodiment of the present invention, the virtual machine 3 to be moved can be appropriately selected.

  The reason for this is to select the virtual machine 3 that behaves closely to the physical machine 2 based on the influence of the fluctuation in the value of the performance information data acquired from both the physical machine 2 and the virtual machine 3.

  Further, according to the first embodiment of the present invention, it is possible to select a migration candidate virtual machine with high accuracy.

  The reason is that the peak timings of the performance information data values of the physical machine 2 and the virtual machine 3 are not coincident with each other, but based on the influence degree of the performance information data values of the physical machine 2 and the virtual machine 3. This is because the migration candidate virtual machine is selected.

  Further, according to the first embodiment of the present invention, it is possible to calculate immediately without using comparative resources.

  This is because the performance information data values of the physical machine 2 and the virtual machine 3 are extracted and evaluated for a predetermined period, respectively, and if the predetermined period is shortened, the amount of data handled can be reduced accordingly. In addition, since immediate past data is used, an immediate calculation result is obtained.

  Further, according to the first embodiment of the present invention, candidate virtual machines 3 can be ranked and presented, so that flexible operation is possible.

  The reason is that the degree of influence is determined for each virtual machine 3 because the degree of influence is calculated for each virtual machine 3 and the degree of influence is calculated according to the magnitude of the degree of influence. .

  The present invention can be applied to a migration virtual machine selection device that selects a migration candidate virtual machine.

DESCRIPTION OF SYMBOLS 1 Virtual machine management apparatus 11 Storage part 12 Selection part 13 Instruction part 14 Collection part 2 Physical machine 3 Virtual machine 4 Virtual machine movement apparatus

Claims (10)

A storage unit that stores a physical machine having one or more virtual machines and performance information of the virtual machines in time series;
A selection unit that selects and outputs one migration candidate virtual machine based on the degree of influence of the performance information in each virtual machine with respect to the performance information in the physical machine;
Only including,
The degree of influence indicates how similar the magnitude of the change in the resource usage of the virtual machine and the magnitude of the change in the resource usage of the physical machine in a predetermined period are.
Virtual machine management device. The virtual machine management apparatus according to claim 1, wherein the influence degree is calculated based on a value obtained by summing a difference between a change amount of the performance information of the physical machine and a change amount of the performance information of one virtual machine. The selection unit selects and outputs the virtual machine having the greatest influence as one movement candidate virtual machine,
The virtual machine management apparatus according to claim 2. The storage unit associates the performance information of the physical machine and the virtual machine with respect to a plurality of types of performance information and stores them in time series,
The selection unit selects a predetermined type of performance information from the performance information based on a predetermined threshold value, and selects and outputs a migration candidate virtual machine using the selected type of performance information. The virtual machine management device according to any one of claims 1 to 3. A physical machine having one or more virtual machines;
A migration candidate machine selection device that selects and outputs a migration candidate virtual machine that is a candidate for migration from the physical machine to another physical machine among the one or more virtual machines;
A virtual machine moving device for moving the movement candidate virtual machine from the physical machine to another physical machine;
Including
The movement candidate machine selection device includes:
A storage unit that stores the performance information of the physical machine and the virtual machine in time series, and
A selection unit that selects and outputs one migration candidate virtual machine based on the degree of influence of the performance information in each virtual machine with respect to the performance information in the physical machine;
A virtual machine movement instruction unit that causes the virtual machine movement device to move the movement candidate virtual machine to another physical machine;
Only including,
The degree of influence indicates how similar the magnitude of the change in the resource usage of the virtual machine and the magnitude of the change in the resource usage of the physical machine in a predetermined period are.
Virtual machine management system. The influence degree is calculated based on a value obtained by summing a difference between a change amount of the performance information of the physical machine and a change amount of the performance information of one virtual machine.
The virtual machine management system according to claim 5. The selection unit selects and outputs the impact is most sizes have a virtual machine as a mobile candidate virtual machine,
The virtual machine management system according to claim 6. Storing a physical machine having one or more virtual machines and performance information of the virtual machines in time series,
Based on the impact of the performance information in each virtual machine on the performance information in the physical machine, select and output one migration candidate virtual machine ,
The degree of influence indicates how similar the magnitude of the change in the resource usage of the virtual machine and the magnitude of the change in the resource usage of the physical machine in a predetermined period are.
Virtual machine management method. The virtual machine management method according to claim 8, wherein the influence degree is calculated based on a value obtained by summing a difference between a change amount of the performance information of the physical machine and a change amount of the performance information of the one virtual machine. If selecting and outputting one of moving candidate virtual machine, the virtual machine management method according to claim 9, wherein the degree of influence selects and outputs a most size have a virtual machine as a mobile candidate virtual machine.

Images