WO2025052509A1 - Estimation device and estimation program - Google Patents

Abstract

An estimation device 10 comprises a workload power consumption estimation unit 103 that links a server power consumption estimation model for learning at least the relationship between a resource use state of a server and power consumption of the server to estimate power consumption on the server, and a workload additional learning model for learning information on power consumption of the workload on the server, and that calculates an estimated value of power consumption when a prescribed workload operates on a server of a movement destination.

Description

Estimation device and estimation program

This disclosure relates to an estimation device and an estimation program.

We want to improve the utilization rate of renewable energy at each data center by moving workloads such as VMs (Virtual Machines) running on servers in the data center to another data center.

To achieve this, it is necessary to estimate the power consumption on a per-workload basis. However, unlike the power consumption of the server itself, the power consumption of a workload cannot be directly measured using a PDU (Power Distribution Unit) or the like.

In this paper, we assume that the power consumption of a server is proportional to the CPU usage rate, and prepare equations for the relationship between each CPU usage rate and the power consumption per 1% of CPU usage rate for multiple servers for the host OS and various VMs. We then estimate the power consumption of the various VMs on the server before the move by calculating the power consumption from these equations. After that, based on the estimation results, we estimate the power consumption of the various VMs on the destination server from the power saving ratio of the server before and after the move.

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

However, even when a workload is moved between servers of the same model and specifications (when the power saving ratio is 1:1), the amount of power decrease on the source server and the amount of power increase on 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.

This disclosure has been made in consideration of the above circumstances, and the purpose of this disclosure is to provide a technology that can estimate the power consumption of a workload with high accuracy.

The estimation device according to one aspect of the present disclosure includes an estimation unit that calculates an estimate of the power consumption of a given workload when it is operated on a destination server by linking a server power consumption estimation model that learns at least the relationship between the resource usage status of the server and the power consumption of the server and an additional learning model for workloads that learns information about the power consumption of the workloads on the server.

The estimation program of one embodiment of the present disclosure causes a computer to function as the estimation device.

This disclosure provides technology that can estimate the power consumption of a workload with high accuracy.

FIG. 1 is a diagram showing an overview of the ideas 1 to 3. FIG. 2 is a diagram showing the load state of the workload. FIG. 3 is a diagram showing the characteristics of the workload. FIG. 4 is a diagram showing the configuration of a system according to this embodiment. FIG. 5 is a diagram showing a process flow of the power consumption estimation method. FIG. 6 is a diagram illustrating a hardware configuration of the estimation device.

Below, an embodiment of the present disclosure will be described with reference to the drawings. In the description of the drawings, the same parts are given the same reference numerals and the description will be omitted.

In this embodiment, the power consumption of a workload is estimated using machine learning. In order to estimate the power consumption with high accuracy in a short time, the following measures 1 to 3 are implemented. Figure 1 shows an overview of measures 1 to 3.

Tip 1:
In this embodiment, the workload power consumption estimation model is considered to be a combination of a server power consumption estimation model and a workload incremental learning model.

The data required to generate a server power consumption estimation model can be obtained in advance in large quantities as actual measurement data from servers, etc. Therefore, in learning phase 1, the relationship between the server resource usage status, etc. and the server's power consumption is learned, and a server power consumption estimation model is generated to estimate the power consumption of the server (including the power consumption of the server's workload) using various information output from the server.

Actual measured data includes, for example, resource usage (CPU usage, memory usage, etc.), workload resource usage, server power consumption, server specifications, server temperature, and temperature surrounding the server.

In learning phase 2, an additional learning model for workloads is generated to learn information about the server, particularly information about the power consumption of the workloads on the server. Specifically, the additional learning model for workloads inputs the estimated value of the power consumption of the workloads on the server estimated by the server power consumption estimation model, the power consumption value of the workload calculated from the actual measured value of the power consumption of the server (hereinafter simply referred to as the actual measured value of the power consumption of the workload), and the characteristic information of the workload (Ingenuity 3 described below) and learns the relationship between them.

For example, the workload additional learning model learns the relationship between the estimated power consumption of a workload and the actual measured power consumption of the workload, the relationship between the estimated power consumption of a workload and the characteristic values of the workload, the relationship between the estimated power consumption of a workload and the actual measured power consumption of the workload and the characteristic values of the workload, the relationship between the estimated power consumption of the same workload on different servers, and the relationship between the estimated power consumption of different workloads on the same server. For example, it learns the error between the estimated value and the actual value due to subtle differences between servers related to manufacturing lots, etc.

In the inference phase, the server power consumption estimation model and the workload additional learning model are linked to estimate the power consumption of the workload on the server. Specifically, the power consumption of a given workload when it runs on the destination server is estimated.

As a method for linking models, for example, the "transfer learning" and "fine tuning" techniques used in the world of natural language processing are used. In other words, in the server power consumption estimation model, among the learning related to the server, the learning related to the workload is entrusted to the workload additional learning model.

When there is learning data such as resource usage that can be obtained in large quantities from a server, and learning data that cannot be obtained in large quantities on a per-workload basis, by using the method described above, it is possible to narrow down the range of model parameter adjustments using training data to the additional learning model for the workload. As a result, it is possible to estimate the power consumption of a workload with high accuracy in a short time.

Tip 2:
As described in Scheme 1, the workload additional learning model inputs the actual measured power consumption of the workload as training data.

However, as shown in Figure 2(a), the load state of a workload generally changes sequentially. Also, as shown in Figure 2(b), it takes a certain amount of time for the power to stabilize in servers A and B before and after the workload transfer. Therefore, it is difficult to determine at what point in time the actual measured power consumption of a server becomes stable before and after the workload transfer.

Therefore, to obtain effective training data (measured values of appropriate power consumption), the server power consumption estimation model estimates the power consumption of the workload on the server, and also estimates the time it takes for the server's power to stabilize (see Figure 1). This time can be estimated, for example, based on measured data on the server's power consumption and measured data on the server's resource usage.

The workload additional learning model uses the actual measured power consumption value after the estimated power stabilization time has elapsed as training data. This increases the value of the actual measured power consumption value as training data, making it possible to estimate the power consumption of a workload with higher accuracy in a short period of time.

Tip 3:
Each workload has its own unique characteristics that affect power consumption, such as the number of threads, availability of GPU acceleration, etc.

For example, workloads that can run on multiple threads will have different power consumption because the number of parallel processes will differ depending on the server's operating environment. For example, the number of threads cannot be increased for workload WL1 shown in Figure 3(a) due to the application's design, but workload WL2 shown in Figure 3(b) can handle four threads. Therefore, even if the destination server is the same, if that server is capable of simultaneously executing four or more threads, workload WL2 will consume more power.

In addition, workloads that can be offloaded to the GPU can be processed in parallel in an environment where the GPU is available in addition to the CPU, which results in differences in power consumption. Depending on how the app is designed, the GPU may not be available or only certain GPUs may be available, resulting in differences in power consumption.

The workload additional learning model uses workload characteristic information as training data (see Figure 1). This makes it possible to absorb differences in power consumption due to workload-specific characteristics, and estimate the power consumption of a workload with even greater accuracy in a short period of time.

FIG. 4 is a diagram showing the configuration of system 1 according to this embodiment.

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 the movement of workloads between servers A and B based on the estimation results. The number of servers may be two or more.

The estimation device 10 includes a resource and power consumption collection unit 101, a server power consumption estimation model generation unit 102, a workload power consumption estimation unit 103, a weighting factor preprocessing unit 104, and a weighting factor adjustment unit 105.

The resource and power consumption collection unit 101 has a function for collecting information on the resource usage status (CPU usage, memory usage, etc.) from servers A and B. For example, the resource and power consumption collection unit 101 collects from server A the CPU usage of the host OS running on server A, and the CPU usage of the workloads running on server A (CPU usage of VM1, CPU usage of VM2, CPU usage of container 1, etc.).

The resource and power consumption collection unit 101 also has a function of collecting information on the power consumption of each server from each PDU of servers A and B. For example, the resource and power consumption collection unit 101 collects the power consumption of server A from the PDU of server A.

The resource and power consumption collection unit 101 has the function of collecting the specification information of each of the servers A and B from the specification information storage DB.

The server power consumption estimation model generation unit 102 has a function of inputting the resource usage status of servers A and B, the power consumption of each server body, each server's specification information, the temperature of each server, the ambient temperature of each server, etc. as learning data, learning the relationship between the resource usage status of servers A and B, the power consumption of each server body, each server's specification information, the temperature of each server, the ambient temperature of each server, etc., and generating a server power consumption estimation model for estimating the power consumption of the server.

Specifically, the server power consumption estimation model estimates the power consumption of the workload on servers A and B from the resource usage of the workload on servers A and B. For example, the server power consumption estimation model estimates the power consumption when a specific workload runs on the destination server. The server power consumption estimation model also estimates the time it will take for the power to stabilize on the destination server after the workload has been moved.

The workload power consumption estimation unit (estimation unit) 103 has a function of inputting the resource usage status of the workload on servers A and B as inference data, linking the server power consumption estimation model with the workload additional learning model, estimating the power consumption of the workload on servers A and B (for example, the power consumption when a specific workload runs on the destination server) from the resource usage status of the workload on servers A and B input as inference data, and estimating the time until the power stabilizes on the destination server after the workload is moved.

Here, we will explain the server power consumption estimation model and the workload additional learning model again.

The server power consumption estimation model works in conjunction with the workload additional learning model, which is used to learn information about the power consumption of workloads on a server, to estimate the power consumption of the workloads running on the server and the time it takes for the server's power to stabilize. As mentioned above, the two models can be linked using techniques such as "transfer learning" and "fine tuning."

The workload additional learning model inputs the estimated value of the workload power consumption on the server estimated by the server power consumption estimation model, the workload power consumption value calculated from the actual measured value of the server power consumption (actual measured value of the workload power consumption) which is the actual measured value after the time has elapsed until the power on the server stabilizes after the workload is moved, and the workload characteristic information stored in the workload characteristic information storage DB, and learns the relationship between them.

When the server power consumption estimation model is not linked to the workload additional learning model, it simply estimates the power consumption of the workload on the server based on the resource usage status of the workload, etc., but when linked to the workload additional learning model, it estimates the power consumption of the workload taking into account the characteristics of the workload and the actual measured value of the power consumption of the transferred workload, etc.

By linking this to an additional learning model for workloads that has been trained using training data such as parameters related to the characteristics of the workload and the actual measured power consumption of the transferred workload, it is possible to estimate the power consumption of the workload with high accuracy in a short time.

The weighting coefficient preprocessing unit 104 has a function of obtaining an estimated time until power becomes stable on the server to which the workload is to be moved from the workload power consumption estimating unit 103, and instructing the resource and power consumption collecting unit 101 to collect resource usage status, etc. after the power stabilization time has elapsed.

The weighting factor adjustment unit 105 has a function of adjusting the weighting factor of the workload additional learning model using the resource usage status and the like collected by the resource and power consumption collection unit 101 as teacher data based on the instructions of the weighting factor preprocessing unit 104. For example, the weighting factor adjustment unit 105 changes the weighting factor of the workload additional learning model so as to reduce the error between the estimated value and the actual measured value of the power consumption of the workload.

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

The execution device 20 also has a function for determining whether or not to move the workload to be moved based on the result of this determination. If the upper limit is not exceeded, it determines that the workload to be moved can be moved and moves it from server A to server B, and if the upper limit is exceeded, it determines that the workload to be moved cannot be moved. Note that this determination method is an example, and other determination methods may be used.

Figure 5 shows the process flow of the power consumption estimation method.

Step S1:
The resource/power consumption collection unit 101 collects the resource usage status (CPU usage, memory usage, etc.) of each server A and B, collects the power consumption of each server body from each PDU of servers A and B, and collects the specification information of each server A and B from a specification information storage DB.

Step S2:
The server power consumption estimation model generation unit 102 learns the relationships between the resource usage status of each server A and B, the power consumption of each server itself, the specification information of each server, the temperature of each server, the ambient temperature of each server, etc., and generates a server power consumption estimation model for estimating the power consumption of the workloads on servers A and B from the resource usage status of the workloads on servers A and B, and for estimating the time it will take for the power to stabilize on the server to which the workload has been moved.

Step S3:
The workload power consumption estimation unit 103 collects the resource usage status of each server A and B, collects the power consumption of each server body from each PDU of servers A and B, and collects the specification information of each server A and B from a specification information storage DB.

Then, the workload power consumption estimation unit 103 links the server power consumption estimation model with an additional learning model for workloads that learns the relationship between the estimated and actual power consumption values of the workload and the characteristic information of the workload, calculates an estimate of the power consumption of the workload to be moved when it runs on the destination server B from the CPU utilization rate of the workload to be moved running on the source server A, and calculates an estimated time until the power on server B stabilizes after the workload to be moved.

Step S4:
The workload power consumption estimator 103 transmits an estimate of the power consumption of the workload to be moved when it operates on the destination server B to the execution device 20. The execution device 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.

Step S5:
The workload power consumption estimator 103 transmits an estimated time until the power of server B stabilizes after the workload to be moved to the weighting factor preprocessor 104. The weighting factor preprocessor 104 instructs the resource and power consumption collector 101 to collect the resource usage status of server B and the power consumption of servers A and B after the estimated time (power stabilization time) has elapsed.

Step S6:
After the workload to be moved is moved from server A to server B and the above-mentioned estimated time has elapsed, the resource/power consumption collection unit 101 collects the resource usage status of the workload to be moved from server B, and collects the power consumption of each server body from each PDU of servers A and B.

Step S7:
The weighting factor adjustment unit 105 adjusts the weighting factor of the workload additional learning model using the resource usage of the workload to be moved on server B collected in step S6 and the actual measured values of power consumption on servers A and B as teacher data.

For example, the weighting factor adjustment unit 105 calculates an optimal value of the weighting factor used in the machine learning of the workload additional learning model so that the difference (error) between the estimated power consumption value estimated in step S3 and the actual power consumption value collected in step S6 is small, and changes the value of the weighting factor to the optimal value. At this time, the workload additional learning model re-learns using the actual power consumption values of servers A and B after the estimated time has elapsed, which were collected in step S6.

By repeating steps S1 to S7, the estimated power consumption of the workload can be improved.

As described above, according to this embodiment, the estimation device 10 is equipped with a workload power consumption estimation unit 103 that links a server power consumption estimation model for estimating the power consumption of a server by learning at least the relationship between the resource usage status of the server and the power consumption of the server, and an additional learning model for workloads for learning information about the power consumption of the workloads on the server, to calculate an estimate of the power consumption of a specified workload when it is operated on a destination server, thereby providing a technology that can estimate the power consumption of a workload with high accuracy.

This disclosure is not limited to the above-described embodiments. Many variations of this disclosure are possible within the scope of the gist of this disclosure.

The estimation device 10 of the present embodiment described above can be realized, for example, as shown in FIG. 6, 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. The memory 902 and the storage 903 are storage devices. In this computer system, the CPU 901 executes a predetermined program loaded onto the memory 902, thereby realizing each function of the estimation device 10.

The estimation device 10 may be implemented in one computer. The estimation device 10 may be implemented in multiple computers. The estimation device 10 may be a virtual machine implemented in a computer.

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.

REFERENCE SIGNS LIST 1 System 10 Estimation device 20 Execution device 101 Resource and power consumption collection unit 102 Server power consumption estimation model generation unit 103 Workload power consumption estimation unit 104 Weighting factor preprocessing unit 105 Weighting factor adjustment unit 901 CPU
902 Memory 903 Storage 904 Communication device 905 Input device 906 Output device

Claims (4)

an estimation unit that calculates an estimate of the power consumption when a predetermined workload is operated on a destination server by linking a server power consumption estimation model for estimating the power consumption of the server by learning at least the relationship between the resource usage status of the server and the power consumption of the server and an additional learning model for workload for learning information on the power consumption of the workload on the server;
An estimation device comprising: The server power consumption estimation model estimates a time until power becomes stable in a destination server to which a workload has been moved;
The workload incremental learning model comprises:
The estimation device according to claim 1 , wherein learning is performed using an actual measurement value of power consumption at the destination server after the estimated power stabilization time has elapsed. The workload additional learning model is trained using workload characteristic information;
The estimation unit is
The estimation device according to claim 1 , wherein the server power consumption estimation model and an additional learning model for the workload are linked to calculate an estimate of power consumption when a predetermined workload is operated on a destination server. An estimation program that causes a computer to function as the estimation device described in claim 1.

Images