Server control method and device, equipment and medium

Abstract

The invention provides a server control method and device, equipment and a medium, and the method comprises the steps: obtaining a plurality of key index monitoring data of a server at different moments; predicting a service load comprehensive index of the server based on the plurality of key index monitoring data in combination with a preset prediction model; and determining an energy consumption adjustment strategy of the server based on the service load comprehensive index, and sending the energy consumption adjustment strategy to the server, so that the server performs energy consumption adjustment according to the energy consumption adjustment strategy. Through the prediction model, the monitoring data of a plurality of key indexes are predicted in real time, the server energy consumption is actively adjusted according to the predicted service load comprehensive indexes on the premise of guaranteeing the normal operation of the service, the passive response is upgraded to active optimization, and the effects of energy conservation, consumption reduction and efficient resource scheduling are achieved.

Description

Technical Field

[0001] The present disclosure relates to the field of computers, and in particular to a server control method, apparatus, device, and medium. Background Art

[0002] In cloud scenarios, hyper-converged platforms need to maintain 24 / 7 uninterrupted operation to ensure business continuity, but business loads exhibit obvious cyclical fluctuations.

[0003] Currently, hyper-converged platforms in related technologies generally adopt a fixed resource configuration model, supplemented by load balancing methods such as virtual machine migration and shutdown. However, fixed configurations are difficult to adapt to dynamic changes in business volume, and can easily lead to resource waste and redundant energy consumption during low business periods. Virtual machine migration and shutdown only provide passive load balancing, and when faced with sudden increases in business volume, they cannot quickly respond to resource demands and suffer significant response delays, making it difficult to meet flexible and efficient business needs. Summary of the Invention

[0004] The present disclosure provides a server control method, apparatus, device and medium, which address the problems in related technologies where fixed resource configuration is used, resulting in idle resources and energy waste during off-peak business periods, and passive load balancing methods such as virtual machine migration or shutdown are unable to quickly respond to sudden increases in business volume.

[0005] The first aspect of the present disclosure proposes a server control method, including: obtaining multiple key indicator monitoring data of the server at different times; based on the multiple key indicator monitoring data, combined with a preset prediction model, predicting the comprehensive business load index of the server; based on the comprehensive business load index, determining the energy consumption adjustment strategy of the server, and sending the energy consumption adjustment strategy to the server, so that the server adjusts energy consumption according to the energy consumption adjustment strategy.

[0006] A second embodiment of the present disclosure provides a server control device, including:

[0007] An acquisition unit is used to obtain multiple key indicator monitoring data of the server at different times;

[0008] A prediction unit is used to predict the comprehensive service load index of the server based on multiple key indicator monitoring data and a preset prediction model;

[0009] The determining unit is used to determine the energy consumption adjustment strategy of the server based on the comprehensive indicator of the business load, and send the energy consumption adjustment strategy to the server so that the server adjusts energy consumption according to the energy consumption adjustment strategy.

[0010] The third aspect embodiment of the present disclosure proposes an electronic device, comprising: a processor and a memory for storing a computer program that can be run on the processor, wherein the processor, when used to run the computer program, executes the method described in the first aspect embodiment of the present disclosure.

[0011] The fourth aspect embodiment of the present disclosure proposes a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable a computer to execute the method described in the first aspect embodiment of the present disclosure.

[0012] The fifth aspect embodiment of the present disclosure provides a computer program product, including a computer program, which implements the method described in the first aspect embodiment of the present disclosure when executed by a processor.

[0013] In summary, according to the server control method provided by the present disclosure, multiple key indicator monitoring data of the server are obtained at different times; based on the multiple key indicator monitoring data, combined with a preset prediction model, a comprehensive business load index of the server is predicted; based on the comprehensive business load index, an energy consumption adjustment strategy for the server is determined, and the energy consumption adjustment strategy is sent to the server so that the server adjusts energy consumption according to the energy consumption adjustment strategy. This method realizes real-time prediction of multiple key indicator monitoring data through the prediction model, and proactively adjusts server energy consumption based on the predicted comprehensive business load index while ensuring normal business operation, upgrading passive response to active optimization, achieving energy saving and consumption reduction and efficient resource scheduling.

[0014] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The accompanying drawings are provided to facilitate a better understanding of the present invention and do not constitute a limitation of the present disclosure.

[0016] Figure 1 A flow chart of a server control method provided by an embodiment of the present disclosure;

[0017] Figure 2 A flowchart of another server control method provided by an embodiment of the present disclosure;

[0018] Figure 3 A schematic diagram of a marker list generation stage provided by an embodiment of the present disclosure;

[0019] Figure 4 A schematic diagram of updating a cluster topology state mapping table provided in an embodiment of the present disclosure;

[0020] Figure 5 A schematic diagram of a specific server control method provided by an embodiment of the present disclosure;

[0021] Figure 6 A schematic structural diagram of a server control device provided in an embodiment of the present disclosure;

[0022] Figure 7 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0023] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0024] Hyper-converged platforms in current cloud scenarios typically require 24 / 7 uninterrupted operation due to business continuity requirements. However, services are not always running at peak performance. While customer scenarios and services vary, business volume generally exhibits certain cyclical usage patterns.

[0025] Hyper-converged platforms in related technologies generally adopt a fixed resource configuration model. This fixed resource configuration model leads to idle system resources and wasted energy during low-peak business periods. Hyper-converged platforms often use the method of migrating or shutting down virtual machines for load balancing. However, shutting down virtual machines cannot be immediately responded to sudden surges in business volume.

[0026] At the same time, based on the rapid development of current artificial intelligence technology (AI technology) and the widespread application of large models, introducing intelligence into the hyper-convergence field has also become a hot topic. The use of AI technology can also upgrade the hyper-convergence system from passive response to active optimization, achieving the goal of energy saving and consumption reduction.

[0027] To address the problems in related technologies, the embodiments of the present disclosure provide a server control method in a hyper-converged platform that dynamically adjusts server energy consumption through intelligent prediction. Specifically, when the hyper-converged system is operating normally, the system monitors the operating status of the system's business volume to obtain multiple key indicator monitoring data of the server at different times; the multiple key indicator monitoring data of the server at different times are sent to the prediction model. After data processing and AI model inference, the system's server energy consumption is actively adjusted while ensuring the normal operation of each business virtual machine.

[0028] The server control method disclosed in the present invention mainly relates to the fields of cloud computing and hyper-convergence, and dynamically adjusts server energy consumption based on intelligent prediction in a hyper-convergence platform.

[0029] The embodiments of the present disclosure will be described in detail below.

[0030] like Figure 1 As shown, an embodiment of the present disclosure provides a server control method, comprising the following steps:

[0031] Step 101: Acquire multiple key indicator monitoring data of the server at different times.

[0032] In some embodiments, the present disclosure can collect core performance data of the server operation in real time through the business monitoring module of the hyper-converged platform, that is, the key indicator monitoring data in the present disclosure.

[0033] In some embodiments, key indicator monitoring data reflects the actual load status of the server at different times and is the basis for subsequent predictions and decisions.

[0034] In some embodiments, key indicator monitoring data is performance data for each key indicator collected at preset time intervals (such as a fixed frequency of every 2 seconds), including key indicator name, value, timestamp and other information, which is used to construct historical data with time series characteristics.

[0035] In some embodiments, key indicators include but are not limited to CPU utilization (reflecting computing resource usage), storage IOPS / throughput (reflecting storage subsystem load), network bandwidth usage (reflecting network transmission pressure), etc., and must cover the three core resources of computing, storage, and network.

[0036] In some embodiments, the server disclosed herein refers to a physical server that serves as a computing node in a hyper-converged platform, and its core regulation object is the CPU (central processing unit) built into the server. By dynamically adjusting parameters such as the CPU frequency and core status (isolation / enablement), refined management of the overall energy consumption of the server is achieved.

[0037] Step 102 : Based on the monitoring data of multiple key indicators and in combination with a preset prediction model, a comprehensive indicator of the server's business load is predicted.

[0038] In some embodiments, the present disclosure may first perform data preprocessing on multiple key indicator monitoring data to obtain smooth target key indicator monitoring data, and then use a preset prediction model (such as a weighted sum model, a machine learning model) to convert the multiple target key indicator monitoring data into a unified comprehensive load indicator (such as a Q value), which can quantitatively reflect the overall business load intensity and trend.

[0039] The comprehensive business load indicator, which can be called the Q value, is a dimensionless value between 0 and 1. A higher value indicates a stronger comprehensive load and can be directly used for subsequent threshold judgment and policy generation.

[0040] Step 103 : determining an energy consumption adjustment strategy for the server based on the comprehensive service load indicator, and sending the energy consumption adjustment strategy to the server so that the server adjusts energy consumption according to the energy consumption adjustment strategy.

[0041] In some embodiments, the present disclosure can determine the indicator duration that meets the adjustment conditions based on the comparison result of the comprehensive indicator value and the preset threshold, and trigger differentiated server energy consumption adjustment strategies based on the indicator duration. After receiving the energy consumption adjustment strategy, the server performs the adjustment action (such as modifying the CPU voltage and frequency curve, updating the core affinity configuration), and feeds back the adjusted status data to the platform, forming an optimization closed loop.

[0042] In summary, according to the server control method proposed in the present disclosure, multiple key indicator monitoring data of the server at different times are obtained; based on the multiple key indicator monitoring data, combined with a preset prediction model, a comprehensive business load index of the server is predicted; based on the comprehensive business load index, an energy consumption adjustment strategy for the server is determined, and the energy consumption adjustment strategy is sent to the server so that the server adjusts energy consumption according to the energy consumption adjustment strategy. This method achieves real-time prediction of multiple key indicator monitoring data through a prediction model, and proactively adjusts server energy consumption based on the predicted comprehensive business load index while ensuring normal business operation, upgrading passive response to active optimization, achieving energy conservation and consumption reduction as well as efficient resource scheduling.

[0043] Figure 2 The flowchart of another server control method proposed in the present disclosure is further shown. Figure 1 The illustrated embodiment further explains, Figure 2 The following steps may be included.

[0044] Step 201: Acquire multiple key indicator monitoring data of the server at different times according to preset time intervals.

[0045] In some embodiments, key indicator monitoring data is obtained from feedback from a service monitoring module. Specifically, the hyperconverged platform uses the service monitoring module to collect key indicator data from servers at different runtimes, including CPU utilization, storage IOPS, and network bandwidth usage, in real time at preset intervals (e.g., every 2 seconds). All data is accurately timestamped, forming a key indicator monitoring data set with time characteristics, providing fundamental support for load analysis and energy consumption regulation.

[0046] For example, during business troughs, even if the overall business volume of the platform's virtual machines is low, the business monitoring module continues to monitor business volume and load data, recording monitoring data of key indicators such as CPU utilization, storage IOPS / throughput, and network bandwidth usage every 2 seconds for subsequent analysis and prediction.

[0047] Step 202 : Determine whether there is any key indicator monitoring data to be processed that is in a target state among the plurality of key indicator monitoring data.

[0048] In some embodiments, it is necessary to screen the status of multiple key indicator monitoring data collected to determine whether there is data to be processed in the target state. The target state in this disclosure includes at least: missing value state (such as data not collected due to monitoring failure), abnormal value state (such as sudden peak or negative value that significantly deviates from business logic), and different value magnitude state (such as the numerical range of each indicator varies greatly due to dimensional differences). By identifying this type of data, a basis is provided for subsequent data cleaning (filling missing values, correcting outliers, and normalization processing), ensuring the data quality used for load analysis and prediction.

[0049] Step 203 : If there is unprocessed key indicator monitoring data in the target state, the unprocessed key indicator monitoring data is preprocessed according to the target state to obtain the preprocessed target key indicator monitoring data.

[0050] In some embodiments, if it is detected that there is data to be processed in the key indicator monitoring data that is in a target state (such as missing values, abnormal values, or different values ​​in magnitude), corresponding preprocessing is performed for different target states.

[0051] Specifically, for key indicator monitoring data with missing values, the average value of the two monitoring values ​​before and after the missing value is taken through linear interpolation to fill in the missing value to maintain the continuity of the time series; for key indicator monitoring data with abnormal values, the 3σ principle (that is, data values ​​outside the range of ±3 times the standard deviation of the mean are considered abnormal) is used to identify abnormal points, and obviously mismeasured abnormal values ​​are directly eliminated, and correctable abnormal values ​​are calibrated with reference to historical trends or adjacent period data; for data with large differences in numerical magnitude due to different dimensions (that is, key indicator monitoring data with different data magnitude states), normalization algorithms are used (such as Min-Max normalization to scale the data to the [0,1] interval, or Z-Score normalization to convert it into a standard normal distribution with a mean of 0 and a standard deviation of 1) to eliminate the dimensional impact between indicators, make different types of data uniformly comparable, and ultimately form smooth and accurate pre-processed data.

[0052] Through the above preprocessing, the present disclosure can obtain smooth, accurate and comparable preprocessed target key indicator monitoring data, providing high-quality input for subsequent load forecasting models.

[0053] Step 204 : Using a preset prediction model and combining the target key indicator monitoring data, a comprehensive indicator of the server's business load is predicted.

[0054] In some embodiments, the performance parameters of the server are determined based on the hardware configuration parameters of the server; the performance parameters and target key monitoring data are input into a preset prediction model to obtain comprehensive indicators of business load at different times.

[0055] Specifically, when the present invention uses a preset prediction model combined with target key indicator monitoring data to predict the comprehensive business load indicators of the server, it first calculates the theoretical performance parameters of the server under ideal conditions (such as the theoretical maximum IOPS of storage and the theoretical maximum bandwidth of the network) through a specific algorithm based on the hardware configuration parameters of the server (such as CPU model, main frequency, disk type, capacity, network card model, rate, etc.); then these performance parameters are used as benchmark values ​​and input into a preset prediction model (such as a weighted formula or a machine learning model) together with the preprocessed target key indicator monitoring data (such as actual CPU utilization, storage IOPS, network bandwidth occupancy, etc.). The prediction model analyzes the correlation between multiple indicators and combines them with their corresponding time to output a comprehensive business load indicator (such as Q value) at different times. This indicator quantitatively reflects the overall load intensity of the server and provides a decision-making basis for subsequent energy consumption adjustment strategies.

[0056] Calculating theoretical server performance parameters under ideal conditions is typically based on specifications provided by the hardware manufacturer and industry-standard algorithms. For example, based on the CPU model, the single-core clock speed (e.g., 3.2GHz), core count (e.g., 8 cores), and architectural features (e.g., hyperthreading) are used to estimate peak computing power using the formula: theoretical computing power = clock speed × number of cores × instructions per cycle (IPC, referencing the architecture's benchmark values). For hard disk drives (HDDs), the theoretical IOPS is calculated based on the rotational speed (e.g., 7200rpm) and average seek time (e.g., 8ms): 1000ms / (average seek time + rotational latency) (rotational latency = 0.5 × (60 / rotational speed)). For solid-state drives (SSDs), the manufacturer's nominal peak sequential / random read / write IOPS is directly used. Based on the network card model (e.g., 10Gbps Ethernet) and transport protocol (e.g., TCP / UDP), after deducting the protocol header overhead (approximately 10%), the theoretical maximum throughput is calculated as nominal rate × (1 - overhead ratio).

[0057] For example, taking CPU utilization, storage IOPS, and network bandwidth as examples, the process of predicting the comprehensive server load indicator is as follows:

[0058] First, when a hyper-converged server is added to the hyper-converged software system, the server's hardware configuration (such as CPU model, main frequency, disk type, capacity, network card model, speed, etc.) will be collected, and the server's performance parameters can be formed through estimated calculations based on the hardware configuration.

[0059] Secondly, the performance parameters and the three indicators of CPU utilization, storage IOPS, and network bandwidth obtained above are pre-processed and sent to the prediction model for key indicator monitoring data. After the prediction model introduces a resource priority weighting mechanism, it realizes the quantitative analysis of multi-dimensional resource correlation.

[0060] The basic formula of the prediction model is: Q = M*C + N*(P / (V*0.75)) + L*(R / (H*0.8)).

[0061] Where C is the actual CPU usage, P is the actual monitored IOPS value, R is the actual network bandwidth, V is the estimated theoretical maximum IOPS value, and H is the estimated maximum network bandwidth. 0.8 and 0.75 in the formula are empirical values, i.e., loss coefficients. M, N, and L are weighted values, and the total weighted value sum is 1.

[0062] Taking the above three indicators as an example, M = 0.8, N = 0.1, L = 0.1, the obtained Q value is less than 1.

[0063] That is, the Q value at each moment is calculated according to Q=0.8*C+0.1*(P / (V*0.75))+0.1*(R / (H*0.8)).

[0064] For example, at 9:38, the CPU usage C value is 70%, the monitored IOPSP value is 4538, the monitored network bandwidth R value is 4.6 Gb / s, V=8000, and H=10 Gb / s.

[0065] Then Q = 0.8*70% + 0.1*(4538 / (8000*0.75)) + 0.1*(4.6 / (10*0.8)) = 0.693;

[0066] For example, at 22:12, the CPU usage C value is 12%, the monitored IOPSP value is 1038, the monitored network bandwidth R value is 1.1 Gb / s, V=8000, and H=10 Gb / s.

[0067] Then Q = 0.8*12% + 0.1*(1038 / (8000*0.75)) + 0.1*(1.1 / (10*0.8)) = 0.127;

[0068] Finally, the Q value of the server at the corresponding moment is obtained (sampling frequency 0.5Hz, once every 2S).

[0069] The present disclosure can also form a prediction curve based on the Q value at each moment and its corresponding moment.

[0070] Step 205: Compare the comprehensive service load metrics at different moments with a preset load threshold to determine the duration of the metrics that meet the adjustment conditions.

[0071] In some embodiments, the present disclosure can compare the comprehensive service load metrics (such as Q value) at different moments with preset load thresholds (such as the first preset load threshold K1 and the second preset load threshold K2, and K1 < K2) in real time, and identify the metric intervals that continuously meet the threshold conditions through sliding window analysis. For example, when the Q value is continuously lower than K2, start timing and count the duration of this state; if the Q value is always higher than K1 and lower than K2 within the interval, it is determined as a medium-low load interval, and the duration needs to reach a preset duration (such as 20 minutes) to trigger the frequency reduction strategy; if the Q value is lower than K1 and the duration meets the standard, it is determined as a low load interval, and the core isolation strategy is triggered.

[0072] It can be understood that if the metric duration is less than the preset duration, no energy consumption adjustment is performed.

[0073] In this way, by combining threshold judgment and time window filtering, it is ensured that the adjustment strategy is only executed when the load state is stable and has actual energy-saving value, avoiding misoperations caused by short-term fluctuations.

[0074] Step 206: If the metric duration is greater than the preset duration, determine the energy consumption adjustment strategy corresponding to the adjustment conditions.

[0075] In the present disclosure, the energy consumption adjustment strategy includes an adjustment time and an adjustment mode. The adjustment time is the same as the preset duration, and the adjustment mode includes a server frequency reduction mode and a server isolation mode.

[0076] In some embodiments, determining the energy consumption adjustment strategy corresponding to the adjustment conditions includes: if the comprehensive service load metric under the first metric duration is greater than or equal to the first preset load threshold and the comprehensive service load metric under the first metric duration is less than the second preset load threshold, determine the adjustment time of the server as the first metric duration and the adjustment mode of the server as the server frequency reduction mode; if the comprehensive service load metric under the first metric duration is less than the first preset load threshold, determine the adjustment time of the server as the first metric duration and the adjustment mode of the server as the server isolation mode.

[0077] In some embodiments, the energy consumption adjustment strategy is determined by comparing the comprehensive service load index (such as the Q value) with preset load thresholds (the first preset load threshold K1, the second preset load threshold K2, and K1 < K2) and a preset duration (such as 20 minutes): If the Q value continuously falls within the interval [K1, K2) (that is, greater than or equal to K1 and less than K2) within a certain time period, and the duration of this first index exceeds the preset duration, it is determined as a medium - low load state. The adjustment time is set as this continuous time period, and the adjustment mode adopts the server down - frequency mode to achieve energy saving by reducing the CPU operating frequency.

[0078] If the Q value continuously remains lower than K1 and the duration of the first index meets the standard, it is determined as a low load state. The adjustment time continues to use this continuous time period, and the adjustment mode switches to the server isolation mode to reduce energy consumption by deactivating some CPU cores.

[0079] In this disclosure, both strategies are triggered on the premise of the stability of the load state (the duration meets the standard) to ensure that the adjustment actions accurately match the service load requirements.

[0080] In summary, according to the server control method proposed in this disclosure, compared with the energy - saving adjustment methods in the related art that rely on periodic tasks or specific times (such as migrating / closing virtual machines during business troughs and回迁 / 开启 during business peaks), the server control method proposed in this disclosure focuses on the hyper - converged platform and innovatively adopts a dynamic CPU energy consumption regulation scheme based on intelligent prediction. This method deploys an intelligent module based on hardware entities in the hyper - converged platform, uses multiple monitoring data collected by time series (such as CPU utilization, storage IOPS, network bandwidth) for periodic training, and continuously improves the accuracy of intelligent prediction. Specifically, the intelligent prediction module can accurately infer the business volume adjustment window period: during the business trough period, actively trigger low - power consumption strategies such as reducing the CPU frequency of virtual machines and fixing virtual machines to run on specific CPU cores (CPU isolation); and before the business peak arrives, lift the CPU isolation and low - power consumption state in advance and dynamically adjust resource configurations and hardware parameters. Through this "prediction, decision - making, execution" closed - loop mechanism, this method can not only ensure business continuity but also achieve smooth adjustment of system power consumption, providing a more efficient and intelligent solution for energy - saving management of the hyper - converged platform.

[0081] Based on Figures 1 to 2 the embodiments shown, as Figure 3 shown, this disclosure provides a schematic diagram of the hardware architecture of a hyper - converged intelligent platform.

[0082] Referring to Figure 3 , the hardware architecture of the hyper - converged intelligent platform consists of multiple hyper - converged servers (for example, 3 hyper - converged servers) and a GPU server (GPU is Graphics Processing Unit, graphics processor).

[0083] The NIC / SSD / HDD (NIC: Network Interface Card; SSD: Solid State Drive; HDD: Hard Disk Drive) and some CPU / MEM resources of multiple hyper-converged servers are used to build a hyper-converged cluster. This construction requires meeting the minimum hyper-converged hardware requirements, such as two SSD cache disks, at least 64GB of memory, and a 10G network card.

[0084] GPU servers are used for building AI models (Artificial Intelligence) and other intelligent computing. You can directly deploy the operating system and build the AI ​​model on the server, or you can first deploy a layer of virtualization software, use a direct passthrough method to connect the GPU to the virtual machine, and then build the AI ​​model. This facilitates the use of AI operations such as intelligent prediction and intelligent reasoning, and matches the appropriate model based on the GPU's video memory, computing power, etc.

[0085] Further, such as Figure 4 As shown, the present disclosure provides a schematic diagram of the software architecture of a hyper-converged intelligent platform. Corresponding to the hardware architecture, the CPU energy consumption regulation, the monitoring of the hyper-converged platform, the hyper-converged platform and the business virtual machines are running on Figure 3 On the hyper-converged server, the scheduling of the intelligent prediction module and the system execution module is placed on the GPU server.

[0086] like Figure 5 As shown, the present disclosure provides a specific server control process diagram and proposes a solution for dynamically adjusting CPU energy consumption based on intelligent prediction in a hyper-converged platform.

[0087] Specifically, during the normal operation of the hyper-converged platform, the business monitoring module in the hyper-converged server monitors the system's business volume, load data, etc. in real time, such as the CPU utilization of the virtual machine, storage IOPS / throughput, network bandwidth occupancy and other indicators. For example, the monitoring data of each indicator is recorded every 2 seconds to obtain multiple key indicator monitoring data. After that, the platform will pre-process the key indicator monitoring data based on the target state of the key indicator monitoring data (for example, fill in missing values, identify and process outliers, and perform normalization) to make the key indicator monitoring data as smooth as possible. The pre-processed target key indicator monitoring data is then transmitted to the intelligent prediction module in the GPU server to predict the comprehensive business load indicators, form a refined comprehensive business load indicator prediction curve, and formulate the adjustment time and adjustment mode of each server in the hyper-converged cluster based on the comprehensive business load indicators. Finally, the server will perform the relevant CPU energy consumption adjustment actions according to the adjustment time and adjustment mode pushed by the intelligent prediction module. The CPU frequency can be increased or decreased according to the energy consumption adjustment strategy, and the CPU mode can be isolated or released.

[0088] In order to implement the server control method provided by the embodiment of the present disclosure, the embodiment of the present disclosure also provides a server control device, such as Figure 6 As shown, the server control device 600 includes:

[0089] An acquisition unit 610 is configured to acquire a plurality of key indicator monitoring data of the server at different times;

[0090] The prediction unit 620 is configured to predict a comprehensive service load indicator of the server based on multiple key indicator monitoring data and a preset prediction model;

[0091] The determining unit 630 is configured to determine an energy consumption adjustment strategy for the server based on the comprehensive service load indicator, and send the energy consumption adjustment strategy to the server so that the server adjusts energy consumption according to the energy consumption adjustment strategy.

[0092] In some embodiments, the acquisition unit 610 is configured to acquire a plurality of key indicator monitoring data of the server at different times according to a preset time interval.

[0093] In some embodiments, the prediction unit 620 is used to: determine whether there is unprocessed key indicator monitoring data in a target state among multiple key indicator monitoring data; if there is unprocessed key indicator monitoring data in a target state, then according to the target state of the unprocessed key indicator monitoring data, the unprocessed key indicator monitoring data is preprocessed corresponding to the target state to obtain the preprocessed target key indicator monitoring data; use a preset prediction model, combined with the target key indicator monitoring data, to predict the comprehensive business load index of the server.

[0094] In some embodiments, the target state includes at least one of a value missing state, a value abnormal state, and a value magnitude different state.

[0095] In some embodiments, the prediction unit 620 is used to: determine the performance parameters of the server based on the hardware configuration parameters of the server; input the performance parameters and target key monitoring data into a preset prediction model to obtain comprehensive indicators of business load at different times.

[0096] In some embodiments, the determination unit 630 is used to: compare the comprehensive business load indicator at different times with the preset load threshold to determine the indicator duration that meets the adjustment conditions; if the indicator duration is greater than the preset duration, then determine the energy consumption adjustment strategy corresponding to the adjustment condition, the energy consumption adjustment strategy includes adjustment time and adjustment mode, the adjustment time is the same as the preset duration, and the adjustment mode includes server frequency reduction mode and server isolation mode.

[0097] In some embodiments, the determination unit 630 is used to: if the comprehensive business load indicator under the first indicator duration is greater than or equal to the first preset load threshold, and the comprehensive business load indicator under the first indicator duration is less than the second preset load threshold, determine that the adjustment time of the server is the first indicator duration, and the adjustment mode of the server is the server frequency reduction mode; if the comprehensive business load indicator under the first indicator duration is less than the first preset load threshold, determine that the adjustment time of the server is the first indicator duration, and the adjustment mode of the server is the server isolation mode.

[0098] It should be noted that the server control device provided in the above embodiment only uses the division of the above-mentioned program modules as an example to illustrate server control. In actual applications, the above-mentioned processing can be assigned to different program modules as needed, that is, the internal structure of the server control device can be divided into different program modules to complete all or part of the processing described above. In addition, the server control device provided in the above embodiment and the server control method embodiment provided in the embodiment of the present disclosure are based on the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0099] Figure 7 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present disclosure is shown in FIG. Figure 7 As shown, the electronic device 700 includes at least one processor 702; and a memory 701 communicatively connected to the at least one processor 702; wherein the memory 701 stores instructions that can be executed by the at least one processor 702, and the instructions are executed by the at least one processor 702 to implement the steps of the server control method described in the embodiment of the present disclosure.

[0100] Optionally, the electronic device may specifically be the server control device of the embodiment of the present disclosure, and the electronic device may implement the corresponding processes implemented by the server control device in each method of the embodiment of the present disclosure. For the sake of brevity, they will not be repeated here.

[0101] It is understood that the electronic device also includes a communication interface 703. The various components in the electronic device are coupled together through a bus system 704. It is understood that the bus system 704 is used to achieve connection and communication between these components. In addition to the data bus, the bus system 704 also includes a power bus, a control bus, and a status signal bus. However, for the sake of clarity, Figure 7 Various buses are labeled as bus system 704 .

[0102] It is understood that the memory 701 can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a flash memory, a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM); the magnetic surface memory can be a magnetic disk memory or a magnetic tape memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), synchronous static random access memory (SSRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct rambus random access memory (DRRAM).The memory 701 described in the embodiments of the present disclosure is intended to include, but is not limited to, these and any other suitable types of memories.

[0103] The methods disclosed in the above embodiments of the present disclosure can be applied to or implemented by processor 702. Processor 702 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by hardware integrated logic circuits in processor 702 or by software instructions. The above processor 702 may be a general-purpose processor, a DSP, or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, etc. Processor 702 can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present disclosure. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in conjunction with the embodiments of the present disclosure can be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module may be located in a storage medium located in memory 701. Processor 702 reads information from memory 701 and, in conjunction with its hardware, completes the steps of the above method.

[0104] In an exemplary embodiment, the electronic device may be implemented by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), FPGAs, general-purpose processors, controllers, MCUs, microprocessors, or other electronic components to perform the aforementioned method.

[0105] The embodiment of the present disclosure further provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable the computer to implement the steps of the server control method described in the embodiment of the present disclosure when executed.

[0106] The embodiment of the present disclosure further provides a computer program product, including a computer program, which implements the steps of the server control method described in the embodiment of the present disclosure when executed by a processor.

[0107] Optionally, the computer-readable storage medium can be applied to the server control device in the embodiment of the present disclosure, and the computer instructions enable the computer to execute the corresponding processes implemented by the server control device in the various methods of the embodiment of the present disclosure. For the sake of brevity, they will not be repeated here.

[0108] In the several embodiments provided in the present disclosure, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0109] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0110] In addition, all functional units in the embodiments of the present disclosure may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0111] Those skilled in the art will understand that all or part of the steps of implementing the above-mentioned method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above-mentioned method embodiment; and the aforementioned storage medium includes: mobile storage devices, ROM, RAM, disks or optical disks, etc. Various media that can store program codes.

[0112] Alternatively, if the above-mentioned integrated unit of the present disclosure is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present disclosure is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present disclosure. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0113] The above description is merely a specific embodiment of the present disclosure, but the scope of protection of the present disclosure is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this disclosure should be included in the scope of protection of the present disclosure. Therefore, the scope of protection of the present disclosure should be based on the scope of protection of the claims.

Claims

1. A server control method, characterized in that: include: Obtain multiple key indicator monitoring data of the server at different times; Based on the multiple key indicator monitoring data and in combination with a preset prediction model, a comprehensive business load indicator of the server is predicted; Based on the comprehensive business load indicator, an energy consumption adjustment strategy for the server is determined, and the energy consumption adjustment strategy is sent to the server, so that the server adjusts energy consumption according to the energy consumption adjustment strategy.

2. The method according to claim 1, characterized in that The acquisition of multiple key indicator monitoring data of the server at different times includes: According to the preset time intervals, multiple key indicator monitoring data of the server at different times are obtained.

3. The method according to claim 1, characterized in that The prediction curve of the server's business load comprehensive indicator prediction based on the multiple key indicator monitoring data and in combination with a preset prediction model includes: Determining whether there is any to-be-processed key indicator monitoring data in a target state among the plurality of key indicator monitoring data; If there is unprocessed key indicator monitoring data in the target state, then according to the target state of the unprocessed key indicator monitoring data, the unprocessed key indicator monitoring data is preprocessed corresponding to the target state to obtain the preprocessed target key indicator monitoring data; The preset prediction model is used in combination with the target key indicator monitoring data to predict the comprehensive business load index of the server.

4. The method according to claim 3, characterized in that The target state includes at least one of a value missing state, a value abnormal state, and a value magnitude different state.

5. The method according to claim 3, characterized in that The use of the preset prediction model in combination with the target key indicator monitoring data to predict the comprehensive business load index of the server includes: Determining performance parameters of the server based on hardware configuration parameters of the server; The performance parameters and the target key monitoring data are input into the preset prediction model to obtain comprehensive indicators of business load at different times.

6. The method according to claim 1, characterized in that The determining of the energy consumption adjustment strategy of the server based on the comprehensive business load indicator includes: Compare the comprehensive business load indicators at different times with the preset load thresholds to determine the indicator duration that meets the adjustment conditions; If the indicator duration is greater than the preset duration, the energy consumption adjustment strategy corresponding to the adjustment condition is determined, and the energy consumption adjustment strategy includes adjustment time and adjustment mode. The adjustment time is the same as the preset duration, and the adjustment mode includes server frequency reduction mode and server isolation mode.

7. The method according to claim 6, characterized in that Determining the energy consumption adjustment strategy corresponding to the adjustment condition includes: If the comprehensive business load indicator during the first indicator duration is greater than or equal to the first preset load threshold, and the comprehensive business load indicator during the first indicator duration is less than the second preset load threshold, determining that the adjustment time of the server is the first indicator duration, and the adjustment mode of the server is the server frequency reduction mode; If the comprehensive business load indicator under the first indicator duration is less than the first preset load threshold, the adjustment time of the server is determined to be the first indicator duration, and the adjustment mode of the server is the server isolation mode.

8. A server control device, characterized in that: The device comprises: An acquisition unit is used to obtain multiple key indicator monitoring data of the server at different times; A prediction unit, configured to predict a comprehensive service load indicator of the server based on the plurality of key indicator monitoring data and in combination with a preset prediction model; A determining unit is used to determine an energy consumption adjustment strategy for the server based on the comprehensive business load indicator, and send the energy consumption adjustment strategy to the server so that the server adjusts energy consumption according to the energy consumption adjustment strategy.

9. An electronic device, characterized in that: include: a processor and a memory for storing a computer program capable of being executed on the processor, Wherein, when the processor is used to run the computer program, it executes the server control method according to any one of claims 1-7.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to enable the computer to execute the server control method according to any one of claims 1 to 7.

Images