Estimation device and estimation program

Abstract

An estimation device 10 comprises: a power consumption estimation unit 104 that uses an estimation model for estimating the power consumption of a workload to calculate an estimated value of the power consumption when a designated workload runs on a migration destination server; and a weighting coefficient adjustment unit 106 that adjusts weighting coefficients of the estimation model on the basis of the estimated value of the power consumption and an actually measured value of the power consumption when the designated workload ran on the migration destination server.

Description

Estimation device and estimation program

[0001] The present disclosure relates to an estimation device and an estimation program.

[0002] By migrating workloads such as virtual machines (VMs) running on servers in a data center to another data center, we hope to improve the renewable energy utilization rate at each data center.

[0003] To achieve this, it is necessary to estimate the power consumption for each workload. However, unlike the power consumption of the server itself, the power consumption of a workload cannot be directly measured using a power distribution unit (PDU) or the like.

[0004] Therefore, in Non-Patent Document 1, it is assumed that the power consumption of a server is proportional to the CPU usage rate, and a relational expression between each CPU usage rate and the power consumption per 1% of CPU usage rate for the host OS and each VM is prepared for each server, and the power consumption of each VM on the server before migration is estimated by calculating the power consumption from these relational expressions. Then, based on the estimation result, the power consumption of each VM on the destination server is estimated from the power saving ratio of the server before and after the migration.

[0005] Yoshikazu Nakamura and six others, "VM Power Consumption Prediction Method Considering VM Migration," 2022 General Conference, B-14-6, March 15, 2022

[0006] However, even when a workload is moved between servers of the same model and specifications (when the power saving ratio is 1:1), the power decrease at the source server and the power increase at the destination server due to the workload movement may not be the same value. Therefore, the method using the power saving ratio between servers as shown in Non-Patent Document 1 has the problem of low estimation accuracy.

[0007] The present disclosure has been made in consideration of the above circumstances, and an object of the present disclosure is to provide a technology that can estimate the power consumption of a workload with high accuracy.

[0008] An estimation device according to one aspect of the present disclosure includes an estimation unit that uses an estimation model for estimating the power consumption of a workload to calculate an estimated value of power consumption when a specified workload is run on a destination server, and an adjustment unit that adjusts a weighting coefficient of the estimation model based on the estimated power consumption and an actual measured value of power consumption when the specified workload is run on the destination server.

[0009] An estimation program according to an aspect of the present disclosure causes a computer to function as the estimation device.

[0010] According to the present disclosure, it is possible to provide a technique that can estimate the power consumption of a workload with high accuracy.

[0011] Fig. 1 is a diagram showing a power consumption estimation model based on machine learning. Fig. 2 is a diagram showing the estimation model of this embodiment. Fig. 3 is a diagram showing the configuration of a system according to this embodiment. Fig. 4 is a diagram showing the processing flow of a power consumption estimation method. Fig. 5 is a diagram showing the hardware configuration of an estimation device.

[0012] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. In the description of the drawings, the same parts are designated by the same reference numerals and the description thereof will be omitted.

[0013] In this embodiment, the power consumption of a workload is estimated by machine learning.

[0014] FIG. 1 is a diagram showing a model for estimating power consumption using machine learning.

[0015] In the learning phase, the relationship between the workload resource usage (CPU usage, memory usage, etc.) and the server power consumption is learned, and an estimation model is generated to estimate the power consumption of the workload on servers A and B based on the workload resource usage on servers A and B.

[0016] In the inference phase, the estimation model is used to calculate an estimate of the power consumption when the workload to be moved runs on the destination server B based on the resource usage of the workload on the source server A.

[0017] For example, the execution device 20 checks the upper limit of the power consumption of the destination server B, and determines whether the power consumption of server B plus the estimated power consumption of the workload to be moved exceeds the upper limit. If the upper limit is not exceeded, the execution device 20 determines that the workload to be moved can be moved, and moves the workload to be moved from server A to server B.

[0018] In this way, since a workload is moved based on the result of estimating the power consumption of the workload, it is possible to obtain both an estimated value of the power consumption due to the workload movement and an actual measured value of the power consumption of the actually moved workload.

[0019] Therefore, in this embodiment, the actual measured values ​​of the power consumption that has actually been transferred and the errors between the actual measured values ​​and the estimated values ​​are fed back to the estimation model as training data, as shown in Fig. 2. In this way, the actual measured values ​​of the power consumption that has actually been transferred and the errors are used as training data for the estimation model, so that the power consumption of the workload can be estimated with high accuracy using the estimation model.

[0020] FIG. 3 is a diagram showing the configuration of the system 1 according to this embodiment.

[0021] The system 1 includes an estimation device 10 that estimates the power consumption of workloads (VMs, containers, etc.) on each of two servers A and B, and an execution device 20 that determines and executes migration of workloads between servers A and B based on the estimation results. The number of servers may be two or more.

[0022] The estimation device 10 includes a resource status collection unit 101 , a power consumption collection unit 102 , a specification information collection unit 103 , a power consumption estimation unit 104 , a migration completion confirmation unit 105 , and a weighting factor adjustment unit 106 .

[0023] The resource status collection unit 101 has a function of collecting information on the respective resource usage statuses (CPU usage rate, memory usage rate, etc.) from servers A and B. For example, the resource status collection unit 101 collects from server A the CPU usage rate of the host OS running on server A and the CPU usage rates of the workloads running on server A (CPU usage rate of VM1, CPU usage rate of VM2, CPU usage rate of container 1, etc.).

[0024] The power consumption collection unit 102 has a function of collecting information on the power consumption of each server body from each PDU of the servers A and B. For example, the power consumption collection unit 102 collects the power consumption of the server A from the PDU of the server A.

[0025] The spec information collection unit 103 has a function of collecting spec information of each of the servers A and B from a spec information storage DB.

[0026] In the learning phase, the power consumption estimation unit (estimation unit) 104 inputs the resource usage status of each server A and B, the power consumption of each server itself, specification information of each server, the temperature of each server, the ambient temperature of each server, etc. as learning data, learns the relationship between the resource usage status of each server A and B, the power consumption of each server itself, specification information of each server, the temperature of each server, the ambient temperature of each server, etc., and has the function of generating an estimation model for estimating the power consumption of the workload on servers A and B from the resource usage status of the workload on servers A and B.

[0027] In the inference phase, the power consumption estimation unit 104 has the function of inputting the resource usage status of the workload on servers A and B as inference data, and using the generated estimation model, estimating the power consumption of the workload on servers A and B from the resource usage status of the workload on servers A and B that has been input as inference data.

[0028] For example, the estimation model is a machine learning model for estimating the power consumption of a server B, which is the destination of a workload, based on the CPU utilization rate of the workload on the server A, which is the source of the workload. When the CPU utilization rate of a workload running on the server A, which is the source of the workload, is input, the power consumption estimation unit 104 uses the estimation model to calculate an estimate of the power consumption of the workload when it runs on the server B, which is the destination of the workload.

[0029] If the estimation model uses the specification information of server B, the temperature of server B, the ambient temperature of server B, etc. as learning data, it calculates an estimated value of the power consumption corresponding to the current values ​​input as inference data.

[0030] The migration completion confirmation unit 105 has a function of confirming the completion of the migration of the workload from the logs of the servers A and B, etc.

[0031] The weighting coefficient adjustment unit (adjustment unit) 106 has the function of calculating the actual measured value of the increased power consumption in the destination server when a workload is moved from the source server to the destination server, and feeding back the actual measured value of the increased power consumption as training data to the estimation model of the power consumption estimation unit 104.

[0032] Specifically, the weighting factor adjustment unit 106 calculates an optimal value for the weighting factor used in the machine learning of the estimation model based on the actual measured value of the increased power consumption on the destination server B due to the workload transfer and the estimated value of power consumption on the destination server B of the workload transfer, and changes the value of the weighting factor to the optimal value. The optimal value is a value that brings the estimated power consumption value closer to the actual measured power consumption value. It can also be said to be a value that minimizes the difference between the estimated value and the actual measured value (power consumption error).

[0033] The weighting factor adjusting unit 106 may function inside the power consumption estimating unit 104 .

[0034] The execution device 20 has a function of checking the upper limit of power consumption of the destination server B and determining whether the power consumption of server B, plus the estimated power consumption of the workload to be moved, exceeds the upper limit.

[0035] The execution device 20 also has a function for determining whether or not to migrate the workload to be migrated based on the determination result. If the upper limit value is not exceeded, the execution device 20 determines that the workload to be migrated is migrated and migrates it from server A to server B. If the upper limit value is exceeded, the execution device 20 determines that the workload to be migrated is migrated. Note that this determination method is an example, and other determination methods may be used.

[0036] FIG. 4 is a diagram showing a processing flow of the power consumption estimation method.

[0037] Step S1: The resource status collection unit 101 collects resource usage status (CPU usage rate, memory usage rate, etc.) from each of servers A and B. The power consumption collection unit 102 collects the power consumption of each of the servers A and B from each PDU of each of servers A and B. The specification information collection unit 103 collects specification information of each of servers A and B from a specification information storage DB.

[0038] Step S2: The resource status collection unit 101 passes the collected resource usage status of servers A and B to the power consumption estimation unit 104. The power consumption collection unit 102 passes the collected power consumption of servers A and B to the power consumption estimation unit 104. The specification information collection unit 103 passes the collected specification information of servers A and B to the power consumption estimation unit 104. Using the generated estimation model, the power consumption estimation unit 104 calculates an estimate of the power consumption of the workload to be moved when it runs on the destination server B, from the CPU usage rate of the workload to be moved running on the source server A.

[0039] Step S3: The power consumption estimating unit 104 transmits an estimate of the power consumption of the workload to be moved when it operates on the destination server B to the execution unit 20. The execution unit 20 determines whether the workload to be moved can be moved to server B, and if it determines that the workload can be moved, moves the workload to be moved from server A to server B.

[0040] Step S4: The migration completion confirmation unit 105 determines from the logs of the servers A and B that the migration of the workload to be migrated has been completed.

[0041] Step S5: The power consumption collection unit 102 collects the power consumption of each server body of servers A and B after the workload is moved from each PDU of servers A and B, and calculates the actual measured value of power consumption at server B, the destination of the workload to be moved, by subtracting the power consumption before the workload is moved from the collected power consumption after the workload is moved.

[0042] Step S6: The weighting coefficient adjustment unit 106 calculates the optimal value of the weighting coefficient used in the machine learning of the estimation model from the difference (error) between the estimated power consumption value estimated in step S2 and the actual measured power consumption value calculated in step S5.

[0043] Step S7: The weighting factor adjustment unit 106 reflects the calculated optimum value in the weighting factor of the estimation model.

[0044] By repeating steps S1 to S7, the estimation result of the power consumption of the workload can be improved.

[0045] As described above, according to this embodiment, the estimation device 10 includes a power consumption estimation unit 104 that uses an estimation model for estimating the power consumption of a workload to calculate an estimated value of the power consumption when a specified workload runs on a destination server, and a weighting coefficient adjustment unit 106 that adjusts the weighting coefficient of the estimation model based on the estimated power consumption and the actual measured value of the power consumption when the specified workload runs on the destination server, thereby providing a technology that can estimate the power consumption of a workload with high accuracy.

[0046] The present disclosure is not limited to the above-described embodiments, and various modifications are possible within the scope of the present disclosure.

[0047] The estimation device 10 of the present embodiment described above can be realized, for example, by using a general-purpose computer system including a CPU 901, a memory 902, a storage 903, a communication device 904, an input device 905, and an output device 906, as shown in Fig. 5. The memory 902 and the storage 903 are storage devices. In the computer system, the CPU 901 executes a predetermined program loaded onto the memory 902, thereby realizing each function of the estimation device 10.

[0048] The estimation device 10 may be implemented by one computer, or by multiple computers, or may be a virtual machine implemented on a computer.

[0049] The program for the estimation device 10 can be stored in a computer-readable recording medium such as a HDD, SSD, USB memory, CD, or DVD. The computer-readable recording medium is, for example, a non-transitory recording medium. The program for the estimation device 10 can also be distributed via a communication network.

[0050] REFERENCE SIGNS LIST 1 System 10 Estimation device 20 Execution device 101 Resource status collection unit 102 Power consumption collection unit 103 Specification information collection unit 104 Power consumption estimation unit 105 Movement completion confirmation unit 106 Weighting coefficient adjustment unit 901 CPU 902 Memory 903 Storage 904 Communication device 905 Input device 906 Output device

Claims

1. An estimation device comprising: an estimation unit that uses an estimation model for estimating the power consumption of a workload to calculate an estimate of the power consumption when a specified workload is run on a destination server; and an adjustment unit that adjusts a weighting coefficient of the estimation model based on the estimated power consumption and an actual measured value of the power consumption when the specified workload is run on the destination server.

2. The estimation device according to claim 1, wherein the adjustment unit adjusts a weighting coefficient of the estimation model so that the estimated value of power consumption approaches an actual measured value of power consumption.

3. An estimation program that causes a computer to function as the estimation device according to claim 1.

Images